mirrors/postgres
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Update FAQ_DEV.

Browse Source
This commit is contained in:
Bruce Momjian 2001-12-29 03:42:59 +00:00
parent 198152730b
commit e0de8d9821
2 changed files with 710 additions and 693 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

639
doc/FAQ_DEV
View File

@ -9,31 +9,146 @@
postgreSQL Web site, http://www.PostgreSQL.org. postgreSQL Web site, http://www.PostgreSQL.org.
_________________________________________________________________ _________________________________________________________________
Questions General Questions
1) What tools are available for developers? 1.1) How do I get involved in PostgreSQL development?
2) What books are good for developers? 1.2) How do I add a feature or fix a bug?
3) Why do we use palloc() and pfree() to allocate memory? 1.3) How do I download/update the current source tree?
4) Why do we use Node and List to make data structures? 1.4) How do I test my changes?
5) How do I add a feature or fix a bug? 1.5) What tools are available for developers?
6) How do I download/update the current source tree? 1.6) What books are good for developers?
7) How do I test my changes? 1.7) What is configure all about?
7) I just added a field to a structure. What else should I do? 1.8) How do I add a new port?
8) Why are table, column, type, function, view names sometimes 1.9) Why don't we use threads in the backend?
1.10) How are RPM's packaged?
1.11) How are CVS branches handled?
Technical Questions
2.1) How do I efficiently access information in tables from the
backend code?
2.2) Why are table, column, type, function, view names sometimes
referenced as Name or NameData, and sometimes as char *? referenced as Name or NameData, and sometimes as char *?
9) How do I efficiently access information in tables from the backend 2.3) Why do we use Node and List to make data structures?
code? 2.4) I just added a field to a structure. What else should I do?
10) What is elog()? 2.5) Why do we use palloc() and pfree() to allocate memory?
11) What is configure all about? 2.6) What is elog()?
12) How do I add a new port? 2.7) What is CommandCounterIncrement()?
13) What is CommandCounterIncrement()?
14) Why don't we use threads in the backend?
15) How are RPM's packaged?
16) How are CVS branches handled?
17) How do I get involved in PostgreSQL development?
_________________________________________________________________ _________________________________________________________________
1) What tools are available for developers? General Questions
1.1) How go I get involved in PostgreSQL development?
This was written by Lamar Owen:
2001-06-22
What open source development process is used by the PostgreSQL team?
Read HACKERS for six months (or a full release cycle, whichever is
longer). Really. HACKERS _is_the process. The process is not well
documented (AFAIK -- it may be somewhere that I am not aware of) --
and it changes continually.
What development environment (OS, system, compilers, etc) is required
to develop code?
Developers Corner on the website has links to this information. The
distribution tarball itself includes all the extra tools and documents
that go beyond a good Unix-like development environment. In general, a
modern unix with a modern gcc, GNU make or equivalent, autoconf (of a
particular version), and good working knowledge of those tools are
required.
What areas need support?
The TODO list.
You've made the first step, by finding and subscribing to HACKERS.
Once you find an area to look at in the TODO, and have read the
documentation on the internals, etc, then you check out a current
CVS,write what you are going to write (keeping your CVS checkout up to
date in the process), and make up a patch (as a context diff only) and
send to the PATCHES list, prefereably.
Discussion on the patch typically happens here. If the patch adds a
major feature, it would be a good idea to talk about it first on the
HACKERS list, in order to increase the chances of it being accepted,
as well as toavoid duplication of effort. Note that experienced
developers with a proven track record usually get the big jobs -- for
more than one reason. Also note that PostgreSQL is highly portable --
nonportable code will likely be dismissed out of hand.
Once your contributions get accepted, things move from there.
Typically, you would be added as a developer on the list on the
website when one of the other developers recommends it. Membership on
the steering committee is by invitation only, by the other steering
committee members, from what I have gathered watching froma distance.
I make these statements from having watched the process for over two
years.
To see a good example of how one goes about this, search the archives
for the name 'Tom Lane' and see what his first post consisted of, and
where he took things. In particular, note that this hasn't been _that_
long ago -- and his bugfixing and general deep knowledge with this
codebase is legendary. Take a few days to read after him. And pay
special attention to both the sheer quantity as well as the
painstaking quality of his work. Both are in high demand.
1.2) How do I add a feature or fix a bug?
The source code is over 250,000 lines. Many problems/features are
isolated to one specific area of the code. Others require knowledge of
much of the source. If you are confused about where to start, ask the
hackers list, and they will be glad to assess the complexity and give
pointers on where to start.
Another thing to keep in mind is that many fixes and features can be
added with surprisingly little code. I often start by adding code,
then looking at other areas in the code where similar things are done,
and by the time I am finished, the patch is quite small and compact.
When adding code, keep in mind that it should use the existing
facilities in the source, for performance reasons and for simplicity.
Often a review of existing code doing similar things is helpful.
1.3) How do I download/update the current source tree?
There are several ways to obtain the source tree. Occasional
developers can just get the most recent source tree snapshot from
ftp.postgresql.org. For regular developers, you can use CVS. CVS
allows you to download the source tree, then occasionally update your
copy of the source tree with any new changes. Using CVS, you don't
have to download the entire source each time, only the changed files.
Anonymous CVS does not allows developers to update the remote source
tree, though privileged developers can do this. There is a CVS FAQ on
our web site that describes how to use remote CVS. You can also use
CVSup, which has similarly functionality, and is available from
ftp.postgresql.org.
To update the source tree, there are two ways. You can generate a
patch against your current source tree, perhaps using the make_diff
tools mentioned above, and send them to the patches list. They will be
reviewed, and applied in a timely manner. If the patch is major, and
we are in beta testing, the developers may wait for the final release
before applying your patches.
For hard-core developers, Marc(scrappy@postgresql.org) will give you a
Unix shell account on postgresql.org, so you can use CVS to update the
main source tree, or you can ftp your files into your account, patch,
and cvs install the changes directly into the source tree.
1.4) How do I test my changes?
First, use psql to make sure it is working as you expect. Then run
src/test/regress and get the output of src/test/regress/checkresults
with and without your changes, to see that your patch does not change
the regression test in unexpected ways. This practice has saved me
many times. The regression tests test the code in ways I would never
do, and has caught many bugs in my patches. By finding the problems
now, you save yourself a lot of debugging later when things are
broken, and you can't figure out when it happened.
1.5) What tools are available for developers?
Aside from the User documentation mentioned in the regular FAQ, there Aside from the User documentation mentioned in the regular FAQ, there
are several development tools available. First, all the files in the are several development tools available. First, all the files in the
@ -141,7 +256,7 @@
is also a script called unused_oids in pgsql/src/include/catalog that is also a script called unused_oids in pgsql/src/include/catalog that
shows the unused oids. shows the unused oids.
2) What books are good for developers? 1.6) What books are good for developers?
I have four good books, An Introduction to Database Systems, by C.J. I have four good books, An Introduction to Database Systems, by C.J.
Date, Addison, Wesley, A Guide to the SQL Standard, by C.J. Date, et. Date, Addison, Wesley, A Guide to the SQL Standard, by C.J. Date, et.
@ -151,239 +266,7 @@
There is also a database performance site, with a handbook on-line There is also a database performance site, with a handbook on-line
written by Jim Gray at http://www.benchmarkresources.com. written by Jim Gray at http://www.benchmarkresources.com.
3) Why do we use palloc() and pfree() to allocate memory? 1.7) What is configure all about?
palloc() and pfree() are used in place of malloc() and free() because
we automatically free all memory allocated when a transaction
completes. This makes it easier to make sure we free memory that gets
allocated in one place, but only freed much later. There are several
contexts that memory can be allocated in, and this controls when the
allocated memory is automatically freed by the backend.
4) Why do we use Node and List to make data structures?
We do this because this allows a consistent way to pass data inside
the backend in a flexible way. Every node has a NodeTag which
specifies what type of data is inside the Node. Lists are groups of
Nodes chained together as a forward-linked list.
Here are some of the List manipulation commands:
lfirst(i)
return the data at list element i.
lnext(i)
return the next list element after i.
foreach(i, list)
loop through list, assigning each list element to i. It is
important to note that i is a List *, not the data in the List
element. You need to use lfirst(i) to get at the data. Here is
a typical code snipped that loops through a List containing Var
*'s and processes each one:
List *i, *list;
foreach(i, list)
{
Var *var = lfirst(i);
/* process var here */
}
lcons(node, list)
add node to the front of list, or create a new list with node
if list is NIL.
lappend(list, node)
add node to the end of list. This is more expensive that lcons.
nconc(list1, list2)
Concat list2 on to the end of list1.
length(list)
return the length of the list.
nth(i, list)
return the i'th element in list.
lconsi, ...
There are integer versions of these: lconsi, lappendi, nthi.
List's containing integers instead of Node pointers are used to
hold list of relation object id's and other integer quantities.
You can print nodes easily inside gdb. First, to disable output
truncation when you use the gdb print command:
(gdb) set print elements 0
Instead of printing values in gdb format, you can use the next two
commands to print out List, Node, and structure contents in a verbose
format that is easier to understand. List's are unrolled into nodes,
and nodes are printed in detail. The first prints in a short format,
and the second in a long format:
(gdb) call print(any_pointer)
(gdb) call pprint(any_pointer)
The output appears in the postmaster log file, or on your screen if
you are running a backend directly without a postmaster.
5) How do I add a feature or fix a bug?
The source code is over 250,000 lines. Many problems/features are
isolated to one specific area of the code. Others require knowledge of
much of the source. If you are confused about where to start, ask the
hackers list, and they will be glad to assess the complexity and give
pointers on where to start.
Another thing to keep in mind is that many fixes and features can be
added with surprisingly little code. I often start by adding code,
then looking at other areas in the code where similar things are done,
and by the time I am finished, the patch is quite small and compact.
When adding code, keep in mind that it should use the existing
facilities in the source, for performance reasons and for simplicity.
Often a review of existing code doing similar things is helpful.
6) How do I download/update the current source tree?
There are several ways to obtain the source tree. Occasional
developers can just get the most recent source tree snapshot from
ftp.postgresql.org. For regular developers, you can use CVS. CVS
allows you to download the source tree, then occasionally update your
copy of the source tree with any new changes. Using CVS, you don't
have to download the entire source each time, only the changed files.
Anonymous CVS does not allows developers to update the remote source
tree, though privileged developers can do this. There is a CVS FAQ on
our web site that describes how to use remote CVS. You can also use
CVSup, which has similarly functionality, and is available from
ftp.postgresql.org.
To update the source tree, there are two ways. You can generate a
patch against your current source tree, perhaps using the make_diff
tools mentioned above, and send them to the patches list. They will be
reviewed, and applied in a timely manner. If the patch is major, and
we are in beta testing, the developers may wait for the final release
before applying your patches.
For hard-core developers, Marc(scrappy@postgresql.org) will give you a
Unix shell account on postgresql.org, so you can use CVS to update the
main source tree, or you can ftp your files into your account, patch,
and cvs install the changes directly into the source tree.
6) How do I test my changes?
First, use psql to make sure it is working as you expect. Then run
src/test/regress and get the output of src/test/regress/checkresults
with and without your changes, to see that your patch does not change
the regression test in unexpected ways. This practice has saved me
many times. The regression tests test the code in ways I would never
do, and has caught many bugs in my patches. By finding the problems
now, you save yourself a lot of debugging later when things are
broken, and you can't figure out when it happened.
7) I just added a field to a structure. What else should I do?
The structures passing around from the parser, rewrite, optimizer, and
executor require quite a bit of support. Most structures have support
routines in src/backend/nodes used to create, copy, read, and output
those structures. Make sure you add support for your new field to
these files. Find any other places the structure may need code for
your new field. mkid is helpful with this (see above).
8) Why are table, column, type, function, view names sometimes referenced as
Name or NameData, and sometimes as char *?
Table, column, type, function, and view names are stored in system
tables in columns of type Name. Name is a fixed-length,
null-terminated type of NAMEDATALEN bytes. (The default value for
NAMEDATALEN is 32 bytes.)
typedef struct nameData
{
char data[NAMEDATALEN];
} NameData;
typedef NameData *Name;
Table, column, type, function, and view names that come into the
backend via user queries are stored as variable-length,
null-terminated character strings.
Many functions are called with both types of names, ie. heap_open().
Because the Name type is null-terminated, it is safe to pass it to a
function expecting a char *. Because there are many cases where
on-disk names(Name) are compared to user-supplied names(char *), there
are many cases where Name and char * are used interchangeably.
9) How do I efficiently access information in tables from the backend code?
You first need to find the tuples(rows) you are interested in. There
are two ways. First, SearchSysCache() and related functions allow you
to query the system catalogs. This is the preferred way to access
system tables, because the first call to the cache loads the needed
rows, and future requests can return the results without accessing the
base table. The caches use system table indexes to look up tuples. A
list of available caches is located in
src/backend/utils/cache/syscache.c.
src/backend/utils/cache/lsyscache.c contains many column-specific
cache lookup functions.
The rows returned are cache-owned versions of the heap rows.
Therefore, you must not modify or delete the tuple returned by
SearchSysCache(). What you should do is release it with
ReleaseSysCache() when you are done using it; this informs the cache
that it can discard that tuple if necessary. If you neglect to call
ReleaseSysCache(), then the cache entry will remain locked in the
cache until end of transaction, which is tolerable but not very
desirable.
If you can't use the system cache, you will need to retrieve the data
directly from the heap table, using the buffer cache that is shared by
all backends. The backend automatically takes care of loading the rows
into the buffer cache.
Open the table with heap_open(). You can then start a table scan with
heap_beginscan(), then use heap_getnext() and continue as long as
HeapTupleIsValid() returns true. Then do a heap_endscan(). Keys can be
assigned to the scan. No indexes are used, so all rows are going to be
compared to the keys, and only the valid rows returned.
You can also use heap_fetch() to fetch rows by block number/offset.
While scans automatically lock/unlock rows from the buffer cache, with
heap_fetch(), you must pass a Buffer pointer, and ReleaseBuffer() it
when completed.
Once you have the row, you can get data that is common to all tuples,
like t_self and t_oid, by merely accessing the HeapTuple structure
entries. If you need a table-specific column, you should take the
HeapTuple pointer, and use the GETSTRUCT() macro to access the
table-specific start of the tuple. You then cast the pointer as a
Form_pg_proc pointer if you are accessing the pg_proc table, or
Form_pg_type if you are accessing pg_type. You can then access the
columns by using a structure pointer:
((Form_pg_class) GETSTRUCT(tuple))->relnatts
You must not directly change live tuples in this way. The best way is
to use heap_modifytuple() and pass it your original tuple, and the
values you want changed. It returns a palloc'ed tuple, which you pass
to heap_replace(). You can delete tuples by passing the tuple's t_self
to heap_destroy(). You use t_self for heap_update() too. Remember,
tuples can be either system cache copies, which may go away after you
call ReleaseSysCache(), or read directly from disk buffers, which go
away when you heap_getnext(), heap_endscan, or ReleaseBuffer(), in the
heap_fetch() case. Or it may be a palloc'ed tuple, that you must
pfree() when finished.
10) What is elog()?
elog() is used to send messages to the front-end, and optionally
terminate the current query being processed. The first parameter is an
elog level of NOTICE, DEBUG, ERROR, or FATAL. NOTICE prints on the
user's terminal and the postmaster logs. DEBUG prints only in the
postmaster logs. ERROR prints in both places, and terminates the
current query, never returning from the call. FATAL terminates the
backend process. The remaining parameters of elog are a printf-style
set of parameters to print.
11) What is configure all about?
The files configure and configure.in are part of the GNU autoconf The files configure and configure.in are part of the GNU autoconf
package. Configure allows us to test for various capabilities of the package. Configure allows us to test for various capabilities of the
@ -405,7 +288,7 @@ typedef struct nameData
removed, so you see only the file contained in the source removed, so you see only the file contained in the source
distribution. distribution.
12) How do I add a new port? 1.8) How do I add a new port?
There are a variety of places that need to be modified to add a new There are a variety of places that need to be modified to add a new
port. First, start in the src/template directory. Add an appropriate port. First, start in the src/template directory. Add an appropriate
@ -422,19 +305,7 @@ typedef struct nameData
src/makefiles directory for port-specific Makefile handling. There is src/makefiles directory for port-specific Makefile handling. There is
a backend/port directory if you need special files for your OS. a backend/port directory if you need special files for your OS.
13) What is CommandCounterIncrement()? 1.9) Why don't we use threads in the backend?
Normally, transactions can not see the rows they modify. This allows
UPDATE foo SET x = x + 1 to work correctly.
However, there are cases where a transactions needs to see rows
affected in previous parts of the transaction. This is accomplished
using a Command Counter. Incrementing the counter allows transactions
to be broken into pieces so each piece can see rows modified by
previous pieces. CommandCounterIncrement() increments the Command
Counter, creating a new part of the transaction.
14) Why don't we use threads in the backend?
There are several reasons threads are not used: There are several reasons threads are not used:
* Historically, threads were unsupported and buggy. * Historically, threads were unsupported and buggy.
@ -443,7 +314,7 @@ typedef struct nameData
remaining backend startup time. remaining backend startup time.
* The backend code would be more complex. * The backend code would be more complex.
15) How are RPM's packaged? 1.10) How are RPM's packaged?
This was written by Lamar Owen: This was written by Lamar Owen:
@ -538,7 +409,7 @@ typedef struct nameData
Of course, there are many projects that DO include all the files Of course, there are many projects that DO include all the files
necessary to build RPMs from their Official Tarball (TM). necessary to build RPMs from their Official Tarball (TM).
16) How are CVS branches managed? 1.11) How are CVS branches managed?
This was written by Tom Lane: This was written by Tom Lane:
@ -597,58 +468,194 @@ typedef struct nameData
tree right away after a major release --- we wait for a dot-release or tree right away after a major release --- we wait for a dot-release or
two, so that we won't have to double-patch the first wave of fixes. two, so that we won't have to double-patch the first wave of fixes.
17) How go I get involved in PostgreSQL development? Technical Questions
2.1) How do I efficiently access information in tables from the backend code?
This was written by Lamar Owen: You first need to find the tuples(rows) you are interested in. There
are two ways. First, SearchSysCache() and related functions allow you
to query the system catalogs. This is the preferred way to access
system tables, because the first call to the cache loads the needed
rows, and future requests can return the results without accessing the
base table. The caches use system table indexes to look up tuples. A
list of available caches is located in
src/backend/utils/cache/syscache.c.
src/backend/utils/cache/lsyscache.c contains many column-specific
cache lookup functions.
2001-06-22 The rows returned are cache-owned versions of the heap rows.
What open source development process is used by the PostgreSQL team? Therefore, you must not modify or delete the tuple returned by
SearchSysCache(). What you should do is release it with
ReleaseSysCache() when you are done using it; this informs the cache
that it can discard that tuple if necessary. If you neglect to call
ReleaseSysCache(), then the cache entry will remain locked in the
cache until end of transaction, which is tolerable but not very
desirable.
Read HACKERS for six months (or a full release cycle, whichever is If you can't use the system cache, you will need to retrieve the data
longer). Really. HACKERS _is_the process. The process is not well directly from the heap table, using the buffer cache that is shared by
documented (AFAIK -- it may be somewhere that I am not aware of) -- all backends. The backend automatically takes care of loading the rows
and it changes continually. into the buffer cache.
What development environment (OS, system, compilers, etc) is required
to develop code?
Developers Corner on the website has links to this information. The Open the table with heap_open(). You can then start a table scan with
distribution tarball itself includes all the extra tools and documents heap_beginscan(), then use heap_getnext() and continue as long as
that go beyond a good Unix-like development environment. In general, a HeapTupleIsValid() returns true. Then do a heap_endscan(). Keys can be
modern unix with a modern gcc, GNU make or equivalent, autoconf (of a assigned to the scan. No indexes are used, so all rows are going to be
particular version), and good working knowledge of those tools are compared to the keys, and only the valid rows returned.
required.
What areas need support?
The TODO list. You can also use heap_fetch() to fetch rows by block number/offset.
While scans automatically lock/unlock rows from the buffer cache, with
heap_fetch(), you must pass a Buffer pointer, and ReleaseBuffer() it
when completed.
You've made the first step, by finding and subscribing to HACKERS. Once you have the row, you can get data that is common to all tuples,
Once you find an area to look at in the TODO, and have read the like t_self and t_oid, by merely accessing the HeapTuple structure
documentation on the internals, etc, then you check out a current entries. If you need a table-specific column, you should take the
CVS,write what you are going to write (keeping your CVS checkout up to HeapTuple pointer, and use the GETSTRUCT() macro to access the
date in the process), and make up a patch (as a context diff only) and table-specific start of the tuple. You then cast the pointer as a
send to the PATCHES list, prefereably. Form_pg_proc pointer if you are accessing the pg_proc table, or
Form_pg_type if you are accessing pg_type. You can then access the
columns by using a structure pointer:
((Form_pg_class) GETSTRUCT(tuple))->relnatts
You must not directly change live tuples in this way. The best way is
to use heap_modifytuple() and pass it your original tuple, and the
values you want changed. It returns a palloc'ed tuple, which you pass
to heap_replace(). You can delete tuples by passing the tuple's t_self
to heap_destroy(). You use t_self for heap_update() too. Remember,
tuples can be either system cache copies, which may go away after you
call ReleaseSysCache(), or read directly from disk buffers, which go
away when you heap_getnext(), heap_endscan, or ReleaseBuffer(), in the
heap_fetch() case. Or it may be a palloc'ed tuple, that you must
pfree() when finished.
Discussion on the patch typically happens here. If the patch adds a 2.2) Why are table, column, type, function, view names sometimes referenced
major feature, it would be a good idea to talk about it first on the as Name or NameData, and sometimes as char *?
HACKERS list, in order to increase the chances of it being accepted,
as well as toavoid duplication of effort. Note that experienced Table, column, type, function, and view names are stored in system
developers with a proven track record usually get the big jobs -- for tables in columns of type Name. Name is a fixed-length,
more than one reason. Also note that PostgreSQL is highly portable -- null-terminated type of NAMEDATALEN bytes. (The default value for
nonportable code will likely be dismissed out of hand. NAMEDATALEN is 32 bytes.)
typedef struct nameData
{
char data[NAMEDATALEN];
} NameData;
typedef NameData *Name;
Table, column, type, function, and view names that come into the
backend via user queries are stored as variable-length,
null-terminated character strings.
Once your contributions get accepted, things move from there. Many functions are called with both types of names, ie. heap_open().
Typically, you would be added as a developer on the list on the Because the Name type is null-terminated, it is safe to pass it to a
website when one of the other developers recommends it. Membership on function expecting a char *. Because there are many cases where
the steering committee is by invitation only, by the other steering on-disk names(Name) are compared to user-supplied names(char *), there
committee members, from what I have gathered watching froma distance. are many cases where Name and char * are used interchangeably.
I make these statements from having watched the process for over two 2.3) Why do we use Node and List to make data structures?
years.
We do this because this allows a consistent way to pass data inside
the backend in a flexible way. Every node has a NodeTag which
specifies what type of data is inside the Node. Lists are groups of
Nodes chained together as a forward-linked list.
To see a good example of how one goes about this, search the archives Here are some of the List manipulation commands:
for the name 'Tom Lane' and see what his first post consisted of, and
where he took things. In particular, note that this hasn't been _that_ lfirst(i)
long ago -- and his bugfixing and general deep knowledge with this return the data at list element i.
codebase is legendary. Take a few days to read after him. And pay
special attention to both the sheer quantity as well as the lnext(i)
painstaking quality of his work. Both are in high demand. return the next list element after i.
foreach(i, list)
loop through list, assigning each list element to i. It is
important to note that i is a List *, not the data in the List
element. You need to use lfirst(i) to get at the data. Here is
a typical code snipped that loops through a List containing Var
*'s and processes each one:
List *i, *list;
foreach(i, list)
{
Var *var = lfirst(i);
/* process var here */
}
lcons(node, list)
add node to the front of list, or create a new list with node
if list is NIL.
lappend(list, node)
add node to the end of list. This is more expensive that lcons.
nconc(list1, list2)
Concat list2 on to the end of list1.
length(list)
return the length of the list.
nth(i, list)
return the i'th element in list.
lconsi, ...
There are integer versions of these: lconsi, lappendi, nthi.
List's containing integers instead of Node pointers are used to
hold list of relation object id's and other integer quantities.
You can print nodes easily inside gdb. First, to disable output
truncation when you use the gdb print command:
(gdb) set print elements 0
Instead of printing values in gdb format, you can use the next two
commands to print out List, Node, and structure contents in a verbose
format that is easier to understand. List's are unrolled into nodes,
and nodes are printed in detail. The first prints in a short format,
and the second in a long format:
(gdb) call print(any_pointer)
(gdb) call pprint(any_pointer)
The output appears in the postmaster log file, or on your screen if
you are running a backend directly without a postmaster.
2.4) I just added a field to a structure. What else should I do?
The structures passing around from the parser, rewrite, optimizer, and
executor require quite a bit of support. Most structures have support
routines in src/backend/nodes used to create, copy, read, and output
those structures. Make sure you add support for your new field to
these files. Find any other places the structure may need code for
your new field. mkid is helpful with this (see above).
2.5) Why do we use palloc() and pfree() to allocate memory?
palloc() and pfree() are used in place of malloc() and free() because
we automatically free all memory allocated when a transaction
completes. This makes it easier to make sure we free memory that gets
allocated in one place, but only freed much later. There are several
contexts that memory can be allocated in, and this controls when the
allocated memory is automatically freed by the backend.
2.6) What is elog()?
elog() is used to send messages to the front-end, and optionally
terminate the current query being processed. The first parameter is an
elog level of NOTICE, DEBUG, ERROR, or FATAL. NOTICE prints on the
user's terminal and the postmaster logs. DEBUG prints only in the
postmaster logs. ERROR prints in both places, and terminates the
current query, never returning from the call. FATAL terminates the
backend process. The remaining parameters of elog are a printf-style
set of parameters to print.
2.7) What is CommandCounterIncrement()?
Normally, transactions can not see the rows they modify. This allows
UPDATE foo SET x = x + 1 to work correctly.
However, there are cases where a transactions needs to see rows
affected in previous parts of the transaction. This is accomplished
using a Command Counter. Incrementing the counter allows transactions
to be broken into pieces so each piece can see rows modified by
previous pieces. CommandCounterIncrement() increments the Command
Counter, creating a new part of the transaction.

764
doc/src/FAQ/FAQ_DEV.html
View File

@ -27,39 +27,169 @@
<CENTER> <CENTER>
<H2>Questions</H2> <H2>General Questions</H2>
</CENTER> </CENTER>
<A href="#1">1</A>) What tools are available for developers?<BR> <A href="#1.1">1.1</A>) How do I get involved in PostgreSQL
<A href="#2">2</A>) What books are good for developers?<BR> development?<BR>
<A href="#3">3</A>) Why do we use <I>palloc</I>() and <A href="#1.2">1.2</A>) How do I add a feature or fix a bug?<BR>
<I>pfree</I>() to allocate memory?<BR> <A href="#1.3">1.3</A>) How do I download/update the current source
<A href="#4">4</A>) Why do we use <I>Node</I> and <I>List</I> to
make data structures?<BR>
<A href="#5">5</A>) How do I add a feature or fix a bug?<BR>
<A href="#6">6</A>) How do I download/update the current source
tree?<BR> tree?<BR>
<A href="#7">7</A>) How do I test my changes?<BR> <A href="#1.4">1.4</A>) How do I test my changes?<BR>
<A href="#7">7</A>) I just added a field to a structure. What else <A href="#1.5">1.5</A>) What tools are available for developers?<BR>
should I do?<BR> <A href="#1.6">1.6</A>) What books are good for developers?<BR>
<A href="#8">8</A>) Why are table, column, type, function, view <A href="#1.7">1.7</A>) What is configure all about?<BR>
<A href="#1.8">1.8</A>) How do I add a new port?<BR>
<A href="#1.9">1.9</A>) Why don't we use threads in the backend?<BR>
<A href="#1.10">1.10</A>) How are RPM's packaged?<BR>
<A href="#1.11">1.11</A>) How are CVS branches handled?<BR>
<H2>Technical Questions</H2>
<A href="#2.1">2.1</A>) How do I efficiently access information in
tables from the backend code?<BR>
<A href="#2.2">2.2</A>) Why are table, column, type, function, view
names sometimes referenced as <I>Name</I> or <I>NameData,</I> and names sometimes referenced as <I>Name</I> or <I>NameData,</I> and
sometimes as <I>char *?</I><BR> sometimes as <I>char *?</I><BR>
<A href="#9">9</A>) How do I efficiently access information in <A href="#2.3">2.3</A>) Why do we use <I>Node</I> and <I>List</I> to
tables from the backend code?<BR> make data structures?<BR>
<A href="#10">10</A>) What is elog()?<BR> <A href="#2.4">2.4</A>) I just added a field to a structure. What else
<A href="#11">11</A>) What is configure all about?<BR> should I do?<BR>
<A href="#12">12</A>) How do I add a new port?<BR> <A href="#2.5">2.5</A>) Why do we use <I>palloc</I>() and
<A href="#13">13</A>) What is CommandCounterIncrement()?<BR> <I>pfree</I>() to allocate memory?<BR>
<A href="#14">14</A>) Why don't we use threads in the backend?<BR> <A href="#2.6">2.6</A>) What is elog()?<BR>
<A href="#15">15</A>) How are RPM's packaged?<BR> <A href="#2.7">2.7</A>) What is CommandCounterIncrement()?<BR>
<A href="#16">16</A>) How are CVS branches handled?<BR>
<A href="#17">17</A>) How do I get involved in PostgreSQL
development?<BR>
<BR> <BR>
<HR> <HR>
<H3><A name="1">1</A>) What tools are available for <CENTER>
<H2>General Questions</H2>
</CENTER>
<H3><A name="1.1">1.1</A>) How go I get involved in PostgreSQL
development?</H3>
<P>This was written by Lamar Owen:</P>
<P>2001-06-22</P>
<B>What open source development process is used by the PostgreSQL
team?</B>
<P>Read HACKERS for six months (or a full release cycle, whichever
is longer). Really. HACKERS _is_the process. The process is not
well documented (AFAIK -- it may be somewhere that I am not aware
of) -- and it changes continually.</P>
<B>What development environment (OS, system, compilers, etc) is
required to develop code?</B>
<P><A href="developers.postgresql.org">Developers Corner</A> on the
website has links to this information. The distribution tarball
itself includes all the extra tools and documents that go beyond a
good Unix-like development environment. In general, a modern unix
with a modern gcc, GNU make or equivalent, autoconf (of a
particular version), and good working knowledge of those tools are
required.</P>
<B>What areas need support?</B>
<P>The TODO list.</P>
<P>You've made the first step, by finding and subscribing to
HACKERS. Once you find an area to look at in the TODO, and have
read the documentation on the internals, etc, then you check out a
current CVS,write what you are going to write (keeping your CVS
checkout up to date in the process), and make up a patch (as a
context diff only) and send to the PATCHES list, prefereably.</P>
<P>Discussion on the patch typically happens here. If the patch
adds a major feature, it would be a good idea to talk about it
first on the HACKERS list, in order to increase the chances of it
being accepted, as well as toavoid duplication of effort. Note that
experienced developers with a proven track record usually get the
big jobs -- for more than one reason. Also note that PostgreSQL is
highly portable -- nonportable code will likely be dismissed out of
hand.</P>
<P>Once your contributions get accepted, things move from there.
Typically, you would be added as a developer on the list on the
website when one of the other developers recommends it. Membership
on the steering committee is by invitation only, by the other
steering committee members, from what I have gathered watching
froma distance.</P>
<P>I make these statements from having watched the process for over
two years.</P>
<P>To see a good example of how one goes about this, search the
archives for the name 'Tom Lane' and see what his first post
consisted of, and where he took things. In particular, note that
this hasn't been _that_ long ago -- and his bugfixing and general
deep knowledge with this codebase is legendary. Take a few days to
read after him. And pay special attention to both the sheer
quantity as well as the painstaking quality of his work. Both are
in high demand.</P>
<H3><A name="1.2">1.2</A>) How do I add a feature or fix a bug?</H3>
<P>The source code is over 250,000 lines. Many problems/features
are isolated to one specific area of the code. Others require
knowledge of much of the source. If you are confused about where to
start, ask the hackers list, and they will be glad to assess the
complexity and give pointers on where to start.</P>
<P>Another thing to keep in mind is that many fixes and features
can be added with surprisingly little code. I often start by adding
code, then looking at other areas in the code where similar things
are done, and by the time I am finished, the patch is quite small
and compact.</P>
<P>When adding code, keep in mind that it should use the existing
facilities in the source, for performance reasons and for
simplicity. Often a review of existing code doing similar things is
helpful.</P>
<H3><A name="1.3">1.3</A>) How do I download/update the current source
tree?</H3>
<P>There are several ways to obtain the source tree. Occasional
developers can just get the most recent source tree snapshot from
ftp.postgresql.org. For regular developers, you can use CVS. CVS
allows you to download the source tree, then occasionally update
your copy of the source tree with any new changes. Using CVS, you
don't have to download the entire source each time, only the
changed files. Anonymous CVS does not allows developers to update
the remote source tree, though privileged developers can do this.
There is a CVS FAQ on our web site that describes how to use remote
CVS. You can also use CVSup, which has similarly functionality, and
is available from ftp.postgresql.org.</P>
<P>To update the source tree, there are two ways. You can generate
a patch against your current source tree, perhaps using the
make_diff tools mentioned above, and send them to the patches list.
They will be reviewed, and applied in a timely manner. If the patch
is major, and we are in beta testing, the developers may wait for
the final release before applying your patches.</P>
<P>For hard-core developers, Marc(scrappy@postgresql.org) will give
you a Unix shell account on postgresql.org, so you can use CVS to
update the main source tree, or you can ftp your files into your
account, patch, and cvs install the changes directly into the
source tree.</P>
<H3><A name="1.4">1.4</A>) How do I test my changes?</H3>
<P>First, use <I>psql</I> to make sure it is working as you expect.
Then run <I>src/test/regress</I> and get the output of
<I>src/test/regress/checkresults</I> with and without your changes,
to see that your patch does not change the regression test in
unexpected ways. This practice has saved me many times. The
regression tests test the code in ways I would never do, and has
caught many bugs in my patches. By finding the problems now, you
save yourself a lot of debugging later when things are broken, and
you can't figure out when it happened.</P>
<H3><A name="1.5">1.5</A>) What tools are available for
developers?</H3> developers?</H3>
<P>Aside from the User documentation mentioned in the regular FAQ, <P>Aside from the User documentation mentioned in the regular FAQ,
@ -179,7 +309,7 @@
There is also a script called <I>unused_oids</I> in There is also a script called <I>unused_oids</I> in
<I>pgsql/src/include/catalog</I> that shows the unused oids.</P> <I>pgsql/src/include/catalog</I> that shows the unused oids.</P>
<H3><A name="2">2</A>) What books are good for developers?</H3> <H3><A name="1.6">1.6</A>) What books are good for developers?</H3>
<P>I have four good books, <I>An Introduction to Database <P>I have four good books, <I>An Introduction to Database
Systems,</I> by C.J. Date, Addison, Wesley, <I>A Guide to the SQL Systems,</I> by C.J. Date, Addison, Wesley, <I>A Guide to the SQL
@ -192,288 +322,7 @@
on-line written by Jim Gray at <A href= on-line written by Jim Gray at <A href=
"http://www.benchmarkresources.com">http://www.benchmarkresources.com.</A></P> "http://www.benchmarkresources.com">http://www.benchmarkresources.com.</A></P>
<H3><A name="3">3</A>) Why do we use <I>palloc</I>() and <H3><A name="1.7">1.7</A>) What is configure all about?</H3>
<I>pfree</I>() to allocate memory?</H3>
<P><I>palloc()</I> and <I>pfree()</I> are used in place of malloc()
and free() because we automatically free all memory allocated when
a transaction completes. This makes it easier to make sure we free
memory that gets allocated in one place, but only freed much later.
There are several contexts that memory can be allocated in, and
this controls when the allocated memory is automatically freed by
the backend.</P>
<H3><A name="4">4</A>) Why do we use <I>Node</I> and <I>List</I> to
make data structures?</H3>
<P>We do this because this allows a consistent way to pass data
inside the backend in a flexible way. Every node has a
<I>NodeTag</I> which specifies what type of data is inside the
Node. <I>Lists</I> are groups of <I>Nodes chained together as a
forward-linked list.</I></P>
<P>Here are some of the <I>List</I> manipulation commands:</P>
<BLOCKQUOTE>
<DL>
<DT>lfirst(i)</DT>
<DD>return the data at list element <I>i.</I></DD>
<DT>lnext(i)</DT>
<DD>return the next list element after <I>i.</I></DD>
<DT>foreach(i, list)</DT>
<DD>
loop through <I>list,</I> assigning each list element to
<I>i.</I> It is important to note that <I>i</I> is a List *,
not the data in the <I>List</I> element. You need to use
<I>lfirst(i)</I> to get at the data. Here is a typical code
snipped that loops through a List containing <I>Var *'s</I>
and processes each one:
<PRE>
<CODE>List *i, *list;
foreach(i, list)
{
Var *var = lfirst(i);
/* process var here */
}
</CODE>
</PRE>
</DD>
<DT>lcons(node, list)</DT>
<DD>add <I>node</I> to the front of <I>list,</I> or create a
new list with <I>node</I> if <I>list</I> is <I>NIL.</I></DD>
<DT>lappend(list, node)</DT>
<DD>add <I>node</I> to the end of <I>list.</I> This is more
expensive that lcons.</DD>
<DT>nconc(list1, list2)</DT>
<DD>Concat <I>list2</I> on to the end of <I>list1.</I></DD>
<DT>length(list)</DT>
<DD>return the length of the <I>list.</I></DD>
<DT>nth(i, list)</DT>
<DD>return the <I>i</I>'th element in <I>list.</I></DD>
<DT>lconsi, ...</DT>
<DD>There are integer versions of these: <I>lconsi, lappendi,
nthi.</I> <I>List's</I> containing integers instead of Node
pointers are used to hold list of relation object id's and
other integer quantities.</DD>
</DL>
</BLOCKQUOTE>
You can print nodes easily inside <I>gdb.</I> First, to disable
output truncation when you use the gdb <I>print</I> command:
<PRE>
<CODE>(gdb) set print elements 0
</CODE>
</PRE>
Instead of printing values in gdb format, you can use the next two
commands to print out List, Node, and structure contents in a
verbose format that is easier to understand. List's are unrolled
into nodes, and nodes are printed in detail. The first prints in a
short format, and the second in a long format:
<PRE>
<CODE>(gdb) call print(any_pointer)
(gdb) call pprint(any_pointer)
</CODE>
</PRE>
The output appears in the postmaster log file, or on your screen if
you are running a backend directly without a postmaster.
<H3><A name="5">5</A>) How do I add a feature or fix a bug?</H3>
<P>The source code is over 250,000 lines. Many problems/features
are isolated to one specific area of the code. Others require
knowledge of much of the source. If you are confused about where to
start, ask the hackers list, and they will be glad to assess the
complexity and give pointers on where to start.</P>
<P>Another thing to keep in mind is that many fixes and features
can be added with surprisingly little code. I often start by adding
code, then looking at other areas in the code where similar things
are done, and by the time I am finished, the patch is quite small
and compact.</P>
<P>When adding code, keep in mind that it should use the existing
facilities in the source, for performance reasons and for
simplicity. Often a review of existing code doing similar things is
helpful.</P>
<H3><A name="6">6</A>) How do I download/update the current source
tree?</H3>
<P>There are several ways to obtain the source tree. Occasional
developers can just get the most recent source tree snapshot from
ftp.postgresql.org. For regular developers, you can use CVS. CVS
allows you to download the source tree, then occasionally update
your copy of the source tree with any new changes. Using CVS, you
don't have to download the entire source each time, only the
changed files. Anonymous CVS does not allows developers to update
the remote source tree, though privileged developers can do this.
There is a CVS FAQ on our web site that describes how to use remote
CVS. You can also use CVSup, which has similarly functionality, and
is available from ftp.postgresql.org.</P>
<P>To update the source tree, there are two ways. You can generate
a patch against your current source tree, perhaps using the
make_diff tools mentioned above, and send them to the patches list.
They will be reviewed, and applied in a timely manner. If the patch
is major, and we are in beta testing, the developers may wait for
the final release before applying your patches.</P>
<P>For hard-core developers, Marc(scrappy@postgresql.org) will give
you a Unix shell account on postgresql.org, so you can use CVS to
update the main source tree, or you can ftp your files into your
account, patch, and cvs install the changes directly into the
source tree.</P>
<H3><A name="6">6</A>) How do I test my changes?</H3>
<P>First, use <I>psql</I> to make sure it is working as you expect.
Then run <I>src/test/regress</I> and get the output of
<I>src/test/regress/checkresults</I> with and without your changes,
to see that your patch does not change the regression test in
unexpected ways. This practice has saved me many times. The
regression tests test the code in ways I would never do, and has
caught many bugs in my patches. By finding the problems now, you
save yourself a lot of debugging later when things are broken, and
you can't figure out when it happened.</P>
<H3><A name="7">7</A>) I just added a field to a structure. What
else should I do?</H3>
<P>The structures passing around from the parser, rewrite,
optimizer, and executor require quite a bit of support. Most
structures have support routines in <I>src/backend/nodes</I> used
to create, copy, read, and output those structures. Make sure you
add support for your new field to these files. Find any other
places the structure may need code for your new field. <I>mkid</I>
is helpful with this (see above).</P>
<H3><A name="8">8</A>) Why are table, column, type, function, view
names sometimes referenced as <I>Name</I> or <I>NameData,</I> and
sometimes as <I>char *?</I></H3>
<P>Table, column, type, function, and view names are stored in
system tables in columns of type <I>Name.</I> Name is a
fixed-length, null-terminated type of <I>NAMEDATALEN</I> bytes.
(The default value for NAMEDATALEN is 32 bytes.)</P>
<PRE>
<CODE>typedef struct nameData
{
char data[NAMEDATALEN];
} NameData;
typedef NameData *Name;
</CODE>
</PRE>
Table, column, type, function, and view names that come into the
backend via user queries are stored as variable-length,
null-terminated character strings.
<P>Many functions are called with both types of names, ie.
<I>heap_open().</I> Because the Name type is null-terminated, it is
safe to pass it to a function expecting a char *. Because there are
many cases where on-disk names(Name) are compared to user-supplied
names(char *), there are many cases where Name and char * are used
interchangeably.</P>
<H3><A name="9">9</A>) How do I efficiently access information in
tables from the backend code?</H3>
<P>You first need to find the tuples(rows) you are interested in.
There are two ways. First, <I>SearchSysCache()</I> and related
functions allow you to query the system catalogs. This is the
preferred way to access system tables, because the first call to
the cache loads the needed rows, and future requests can return the
results without accessing the base table. The caches use system
table indexes to look up tuples. A list of available caches is
located in <I>src/backend/utils/cache/syscache.c.</I>
<I>src/backend/utils/cache/lsyscache.c</I> contains many
column-specific cache lookup functions.</P>
<P>The rows returned are cache-owned versions of the heap rows.
Therefore, you must not modify or delete the tuple returned by
<I>SearchSysCache()</I>. What you <I>should</I> do is release it
with <I>ReleaseSysCache()</I> when you are done using it; this
informs the cache that it can discard that tuple if necessary. If
you neglect to call <I>ReleaseSysCache()</I>, then the cache entry
will remain locked in the cache until end of transaction, which is
tolerable but not very desirable.</P>
<P>If you can't use the system cache, you will need to retrieve the
data directly from the heap table, using the buffer cache that is
shared by all backends. The backend automatically takes care of
loading the rows into the buffer cache.</P>
<P>Open the table with <I>heap_open().</I> You can then start a
table scan with <I>heap_beginscan(),</I> then use
<I>heap_getnext()</I> and continue as long as
<I>HeapTupleIsValid()</I> returns true. Then do a
<I>heap_endscan().</I> <I>Keys</I> can be assigned to the
<I>scan.</I> No indexes are used, so all rows are going to be
compared to the keys, and only the valid rows returned.</P>
<P>You can also use <I>heap_fetch()</I> to fetch rows by block
number/offset. While scans automatically lock/unlock rows from the
buffer cache, with <I>heap_fetch(),</I> you must pass a
<I>Buffer</I> pointer, and <I>ReleaseBuffer()</I> it when
completed.</P>
<P>Once you have the row, you can get data that is common to all
tuples, like <I>t_self</I> and <I>t_oid,</I> by merely accessing
the <I>HeapTuple</I> structure entries. If you need a
table-specific column, you should take the HeapTuple pointer, and
use the <I>GETSTRUCT()</I> macro to access the table-specific start
of the tuple. You then cast the pointer as a <I>Form_pg_proc</I>
pointer if you are accessing the pg_proc table, or
<I>Form_pg_type</I> if you are accessing pg_type. You can then
access the columns by using a structure pointer:</P>
<PRE>
<CODE>((Form_pg_class) GETSTRUCT(tuple))-&gt;relnatts
</CODE>
</PRE>
You must not directly change <I>live</I> tuples in this way. The
best way is to use <I>heap_modifytuple()</I> and pass it your
original tuple, and the values you want changed. It returns a
palloc'ed tuple, which you pass to <I>heap_replace().</I> You can
delete tuples by passing the tuple's <I>t_self</I> to
<I>heap_destroy().</I> You use <I>t_self</I> for
<I>heap_update()</I> too. Remember, tuples can be either system
cache copies, which may go away after you call
<I>ReleaseSysCache()</I>, or read directly from disk buffers, which
go away when you <I>heap_getnext()</I>, <I>heap_endscan</I>, or
<I>ReleaseBuffer()</I>, in the <I>heap_fetch()</I> case. Or it may
be a palloc'ed tuple, that you must <I>pfree()</I> when finished.
<H3><A name="10">10</A>) What is elog()?</H3>
<P><I>elog()</I> is used to send messages to the front-end, and
optionally terminate the current query being processed. The first
parameter is an elog level of <I>NOTICE,</I> <I>DEBUG,</I>
<I>ERROR,</I> or <I>FATAL.</I> <I>NOTICE</I> prints on the user's
terminal and the postmaster logs. <I>DEBUG</I> prints only in the
postmaster logs. <I>ERROR</I> prints in both places, and terminates
the current query, never returning from the call. <I>FATAL</I>
terminates the backend process. The remaining parameters of
<I>elog</I> are a <I>printf</I>-style set of parameters to
print.</P>
<H3><A name="11">11</A>) What is configure all about?</H3>
<P>The files <I>configure</I> and <I>configure.in</I> are part of <P>The files <I>configure</I> and <I>configure.in</I> are part of
the GNU <I>autoconf</I> package. Configure allows us to test for the GNU <I>autoconf</I> package. Configure allows us to test for
@ -497,7 +346,7 @@
all files derived by configure are removed, so you see only the all files derived by configure are removed, so you see only the
file contained in the source distribution.</P> file contained in the source distribution.</P>
<H3><A name="12">12</A>) How do I add a new port?</H3> <H3><A name="1.8">1.8</A>) How do I add a new port?</H3>
<P>There are a variety of places that need to be modified to add a <P>There are a variety of places that need to be modified to add a
new port. First, start in the <I>src/template</I> directory. Add an new port. First, start in the <I>src/template</I> directory. Add an
@ -516,20 +365,7 @@
handling. There is a <I>backend/port</I> directory if you need handling. There is a <I>backend/port</I> directory if you need
special files for your OS.</P> special files for your OS.</P>
<H3><A name="13">13</A>) What is CommandCounterIncrement()?</H3> <H3><A name="1.9">1.9</A>) Why don't we use threads in the
<P>Normally, transactions can not see the rows they modify. This
allows <CODE>UPDATE foo SET x = x + 1</CODE> to work correctly.</P>
<P>However, there are cases where a transactions needs to see rows
affected in previous parts of the transaction. This is accomplished
using a Command Counter. Incrementing the counter allows
transactions to be broken into pieces so each piece can see rows
modified by previous pieces. <I>CommandCounterIncrement()</I>
increments the Command Counter, creating a new part of the
transaction.</P>
<H3><A name="14">14</A>) Why don't we use threads in the
backend?</H3> backend?</H3>
<P>There are several reasons threads are not used:</P> <P>There are several reasons threads are not used:</P>
@ -545,7 +381,7 @@
<LI>The backend code would be more complex.</LI> <LI>The backend code would be more complex.</LI>
</UL> </UL>
<H3><A name="15">15</A>) How are RPM's packaged?</H3> <H3><A name="1.10">1.10</A>) How are RPM's packaged?</H3>
<P>This was written by Lamar Owen:</P> <P>This was written by Lamar Owen:</P>
@ -650,7 +486,7 @@
<P>Of course, there are many projects that DO include all the files <P>Of course, there are many projects that DO include all the files
necessary to build RPMs from their Official Tarball (TM).</P> necessary to build RPMs from their Official Tarball (TM).</P>
<H3><A name="16">16</A>) How are CVS branches managed?</H3> <H3><A name="1.11">1.11</A>) How are CVS branches managed?</H3>
<P>This was written by Tom Lane:</P> <P>This was written by Tom Lane:</P>
@ -720,70 +556,244 @@
dot-release or two, so that we won't have to double-patch the first dot-release or two, so that we won't have to double-patch the first
wave of fixes.</P> wave of fixes.</P>
<H3><A name="17">17</A>) How go I get involved in PostgreSQL <CENTER>
development?</H3> <H2>Technical Questions</H2>
</CENTER>
<P>This was written by Lamar Owen:</P> <H3><A name="2.1">2.1</A>) How do I efficiently access information in
tables from the backend code?</H3>
<P>2001-06-22</P> <P>You first need to find the tuples(rows) you are interested in.
There are two ways. First, <I>SearchSysCache()</I> and related
functions allow you to query the system catalogs. This is the
preferred way to access system tables, because the first call to
the cache loads the needed rows, and future requests can return the
results without accessing the base table. The caches use system
table indexes to look up tuples. A list of available caches is
located in <I>src/backend/utils/cache/syscache.c.</I>
<I>src/backend/utils/cache/lsyscache.c</I> contains many
column-specific cache lookup functions.</P>
<B>What open source development process is used by the PostgreSQL <P>The rows returned are cache-owned versions of the heap rows.
team?</B> Therefore, you must not modify or delete the tuple returned by
<I>SearchSysCache()</I>. What you <I>should</I> do is release it
with <I>ReleaseSysCache()</I> when you are done using it; this
informs the cache that it can discard that tuple if necessary. If
you neglect to call <I>ReleaseSysCache()</I>, then the cache entry
will remain locked in the cache until end of transaction, which is
tolerable but not very desirable.</P>
<P>Read HACKERS for six months (or a full release cycle, whichever <P>If you can't use the system cache, you will need to retrieve the
is longer). Really. HACKERS _is_the process. The process is not data directly from the heap table, using the buffer cache that is
well documented (AFAIK -- it may be somewhere that I am not aware shared by all backends. The backend automatically takes care of
of) -- and it changes continually.</P> loading the rows into the buffer cache.</P>
<B>What development environment (OS, system, compilers, etc) is <P>Open the table with <I>heap_open().</I> You can then start a
required to develop code?</B> table scan with <I>heap_beginscan(),</I> then use
<I>heap_getnext()</I> and continue as long as
<I>HeapTupleIsValid()</I> returns true. Then do a
<I>heap_endscan().</I> <I>Keys</I> can be assigned to the
<I>scan.</I> No indexes are used, so all rows are going to be
compared to the keys, and only the valid rows returned.</P>
<P><A href="developers.postgresql.org">Developers Corner</A> on the <P>You can also use <I>heap_fetch()</I> to fetch rows by block
website has links to this information. The distribution tarball number/offset. While scans automatically lock/unlock rows from the
itself includes all the extra tools and documents that go beyond a buffer cache, with <I>heap_fetch(),</I> you must pass a
good Unix-like development environment. In general, a modern unix <I>Buffer</I> pointer, and <I>ReleaseBuffer()</I> it when
with a modern gcc, GNU make or equivalent, autoconf (of a completed.</P>
particular version), and good working knowledge of those tools are
required.</P>
<B>What areas need support?</B> <P>Once you have the row, you can get data that is common to all
tuples, like <I>t_self</I> and <I>t_oid,</I> by merely accessing
the <I>HeapTuple</I> structure entries. If you need a
table-specific column, you should take the HeapTuple pointer, and
use the <I>GETSTRUCT()</I> macro to access the table-specific start
of the tuple. You then cast the pointer as a <I>Form_pg_proc</I>
pointer if you are accessing the pg_proc table, or
<I>Form_pg_type</I> if you are accessing pg_type. You can then
access the columns by using a structure pointer:</P>
<PRE>
<CODE>((Form_pg_class) GETSTRUCT(tuple))-&gt;relnatts
</CODE>
</PRE>
You must not directly change <I>live</I> tuples in this way. The
best way is to use <I>heap_modifytuple()</I> and pass it your
original tuple, and the values you want changed. It returns a
palloc'ed tuple, which you pass to <I>heap_replace().</I> You can
delete tuples by passing the tuple's <I>t_self</I> to
<I>heap_destroy().</I> You use <I>t_self</I> for
<I>heap_update()</I> too. Remember, tuples can be either system
cache copies, which may go away after you call
<I>ReleaseSysCache()</I>, or read directly from disk buffers, which
go away when you <I>heap_getnext()</I>, <I>heap_endscan</I>, or
<I>ReleaseBuffer()</I>, in the <I>heap_fetch()</I> case. Or it may
be a palloc'ed tuple, that you must <I>pfree()</I> when finished.
<P>The TODO list.</P> <H3><A name="2.2">2.2</A>) Why are table, column, type, function, view
names sometimes referenced as <I>Name</I> or <I>NameData,</I> and
sometimes as <I>char *?</I></H3>
<P>You've made the first step, by finding and subscribing to <P>Table, column, type, function, and view names are stored in
HACKERS. Once you find an area to look at in the TODO, and have system tables in columns of type <I>Name.</I> Name is a
read the documentation on the internals, etc, then you check out a fixed-length, null-terminated type of <I>NAMEDATALEN</I> bytes.
current CVS,write what you are going to write (keeping your CVS (The default value for NAMEDATALEN is 32 bytes.)</P>
checkout up to date in the process), and make up a patch (as a <PRE>
context diff only) and send to the PATCHES list, prefereably.</P> <CODE>typedef struct nameData
{
char data[NAMEDATALEN];
} NameData;
typedef NameData *Name;
</CODE>
</PRE>
Table, column, type, function, and view names that come into the
backend via user queries are stored as variable-length,
null-terminated character strings.
<P>Discussion on the patch typically happens here. If the patch <P>Many functions are called with both types of names, ie.
adds a major feature, it would be a good idea to talk about it <I>heap_open().</I> Because the Name type is null-terminated, it is
first on the HACKERS list, in order to increase the chances of it safe to pass it to a function expecting a char *. Because there are
being accepted, as well as toavoid duplication of effort. Note that many cases where on-disk names(Name) are compared to user-supplied
experienced developers with a proven track record usually get the names(char *), there are many cases where Name and char * are used
big jobs -- for more than one reason. Also note that PostgreSQL is interchangeably.</P>
highly portable -- nonportable code will likely be dismissed out of
hand.</P>
<P>Once your contributions get accepted, things move from there. <H3><A name="2.3">2.3</A>) Why do we use <I>Node</I> and <I>List</I> to
Typically, you would be added as a developer on the list on the make data structures?</H3>
website when one of the other developers recommends it. Membership
on the steering committee is by invitation only, by the other
steering committee members, from what I have gathered watching
froma distance.</P>
<P>I make these statements from having watched the process for over <P>We do this because this allows a consistent way to pass data
two years.</P> inside the backend in a flexible way. Every node has a
<I>NodeTag</I> which specifies what type of data is inside the
Node. <I>Lists</I> are groups of <I>Nodes chained together as a
forward-linked list.</I></P>
<P>Here are some of the <I>List</I> manipulation commands:</P>
<BLOCKQUOTE>
<DL>
<DT>lfirst(i)</DT>
<DD>return the data at list element <I>i.</I></DD>
<DT>lnext(i)</DT>
<DD>return the next list element after <I>i.</I></DD>
<DT>foreach(i, list)</DT>
<DD>
loop through <I>list,</I> assigning each list element to
<I>i.</I> It is important to note that <I>i</I> is a List *,
not the data in the <I>List</I> element. You need to use
<I>lfirst(i)</I> to get at the data. Here is a typical code
snipped that loops through a List containing <I>Var *'s</I>
and processes each one:
<PRE>
<CODE>List *i, *list;
foreach(i, list)
{
Var *var = lfirst(i);
/* process var here */
}
</CODE>
</PRE>
</DD>
<DT>lcons(node, list)</DT>
<DD>add <I>node</I> to the front of <I>list,</I> or create a
new list with <I>node</I> if <I>list</I> is <I>NIL.</I></DD>
<DT>lappend(list, node)</DT>
<DD>add <I>node</I> to the end of <I>list.</I> This is more
expensive that lcons.</DD>
<DT>nconc(list1, list2)</DT>
<DD>Concat <I>list2</I> on to the end of <I>list1.</I></DD>
<DT>length(list)</DT>
<DD>return the length of the <I>list.</I></DD>
<DT>nth(i, list)</DT>
<DD>return the <I>i</I>'th element in <I>list.</I></DD>
<DT>lconsi, ...</DT>
<DD>There are integer versions of these: <I>lconsi, lappendi,
nthi.</I> <I>List's</I> containing integers instead of Node
pointers are used to hold list of relation object id's and
other integer quantities.</DD>
</DL>
</BLOCKQUOTE>
You can print nodes easily inside <I>gdb.</I> First, to disable
output truncation when you use the gdb <I>print</I> command:
<PRE>
<CODE>(gdb) set print elements 0
</CODE>
</PRE>
Instead of printing values in gdb format, you can use the next two
commands to print out List, Node, and structure contents in a
verbose format that is easier to understand. List's are unrolled
into nodes, and nodes are printed in detail. The first prints in a
short format, and the second in a long format:
<PRE>
<CODE>(gdb) call print(any_pointer)
(gdb) call pprint(any_pointer)
</CODE>
</PRE>
The output appears in the postmaster log file, or on your screen if
you are running a backend directly without a postmaster.
<H3><A name="2.4">2.4</A>) I just added a field to a structure. What
else should I do?</H3>
<P>The structures passing around from the parser, rewrite,
optimizer, and executor require quite a bit of support. Most
structures have support routines in <I>src/backend/nodes</I> used
to create, copy, read, and output those structures. Make sure you
add support for your new field to these files. Find any other
places the structure may need code for your new field. <I>mkid</I>
is helpful with this (see above).</P>
<H3><A name="2.5">2.5</A>) Why do we use <I>palloc</I>() and
<I>pfree</I>() to allocate memory?</H3>
<P><I>palloc()</I> and <I>pfree()</I> are used in place of malloc()
and free() because we automatically free all memory allocated when
a transaction completes. This makes it easier to make sure we free
memory that gets allocated in one place, but only freed much later.
There are several contexts that memory can be allocated in, and
this controls when the allocated memory is automatically freed by
the backend.</P>
<H3><A name="2.6">2.6</A>) What is elog()?</H3>
<P><I>elog()</I> is used to send messages to the front-end, and
optionally terminate the current query being processed. The first
parameter is an elog level of <I>NOTICE,</I> <I>DEBUG,</I>
<I>ERROR,</I> or <I>FATAL.</I> <I>NOTICE</I> prints on the user's
terminal and the postmaster logs. <I>DEBUG</I> prints only in the
postmaster logs. <I>ERROR</I> prints in both places, and terminates
the current query, never returning from the call. <I>FATAL</I>
terminates the backend process. The remaining parameters of
<I>elog</I> are a <I>printf</I>-style set of parameters to
print.</P>
<H3><A name="2.7">2.7</A>) What is CommandCounterIncrement()?</H3>
<P>Normally, transactions can not see the rows they modify. This
allows <CODE>UPDATE foo SET x = x + 1</CODE> to work correctly.</P>
<P>However, there are cases where a transactions needs to see rows
affected in previous parts of the transaction. This is accomplished
using a Command Counter. Incrementing the counter allows
transactions to be broken into pieces so each piece can see rows
modified by previous pieces. <I>CommandCounterIncrement()</I>
increments the Command Counter, creating a new part of the
transaction.</P>
<P>To see a good example of how one goes about this, search the
archives for the name 'Tom Lane' and see what his first post
consisted of, and where he took things. In particular, note that
this hasn't been _that_ long ago -- and his bugfixing and general
deep knowledge with this codebase is legendary. Take a few days to
read after him. And pay special attention to both the sheer
quantity as well as the painstaking quality of his work. Both are
in high demand.</P>
</BODY> </BODY>
</HTML> </HTML>