Re: [HACKERS] directory archive format for pg_dump
Moving onto the directory archive part of this patch, the feature seems to work as advertised; here's a quick test case: createdb pgbench pgbench -i -s 1 pgbench pg_dump -F d -f test pg_restore -k test pg_restore -l test createdb copy pg_restore -d copy test The copy made that way looked good. There's a good chunk of code in the patch that revolves around BLOB support. We need to get someone who is more familiar with those than me to suggest some tests for that part before this gets committed. If you could suggest how to test that code, that would be helpful. There's a number of small things that I'd like to see improved in new rev of this code pg_dump: help message for --file needs to mention that this is overloaded to also specify the output directory too. pg_dump: the documentation for --file should say the directory is created, and must not exist when you start. The code catches this well, but that expectation is not clear until you try it. pg_restore: the help message check the directory archive would be clearer as check an archive in directory format. There are some tab vs. space whitespace inconsistencies in the documentation added. The comments at the beginning of functions could be more consistent. Early parts of the code have a header for each function that's extensive. Maybe even a bit more than needed. I'm not sure why it's important to document here which of these functions is optional/mandatory for example, and getting rid of just those would trim a decent number of lines out of the patch. But then at the end, all of the new functions added aren't documented at all. Some of those are near trivial, but it would be better to have at least a small descriptive header for them. The comment header at the beginning of pg_backup_directory is a bit weird. I guess Philip Warner should still be credited as the author of the code this was based on, but it's a weird seeing a new file attributed solely to him. Also, there's an XXX in the identification field there that should be filled in with the file name. There's your feedback for this round. I hope we'll see an updated patch from you as part of the next CommitFest. -- Greg Smith 2ndQuadrant USg...@2ndquadrant.com Baltimore, MD PostgreSQL Training, Services and Supportwww.2ndQuadrant.us PostgreSQL 9.0 High Performance: http://www.2ndQuadrant.com/books -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 16.12.2010 12:12, Greg Smith wrote: Moving onto the directory archive part of this patch, the feature seems to work as advertised; here's a quick test case: createdb pgbench pgbench -i -s 1 pgbench pg_dump -F d -f test pg_restore -k test pg_restore -l test createdb copy pg_restore -d copy test The copy made that way looked good. There's a good chunk of code in the patch that revolves around BLOB support. We need to get someone who is more familiar with those than me to suggest some tests for that part before this gets committed. If you could suggest how to test that code, that would be helpful. There's a number of small things that I'd like to see improved in new rev of this code ... In addition to those: The check functionality seems orthogonal, it should be splitted off to a separate patch. It would possibly be useful to be perform sanity checks on an archive in custom format too, and the directory format works just as well without it. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 16.12.2010 17:23, Heikki Linnakangas wrote: On 16.12.2010 12:12, Greg Smith wrote: There's a number of small things that I'd like to see improved in new rev of this code ... In addition to those: ... One more thing: the motivation behind this patch is to allow parallel pg_dump in the future, so we should be make sure this patch caters well for that. As soon as we have parallel pg_dump, the next big thing is going to be parallel dump of the same table using multiple processes. Perhaps we should prepare for that in the directory archive format, by allowing the data of a single table to be split into multiple files. That way parallel pg_dump is simple, you just split the table in chunks of roughly the same size, say 10GB each, and launch a process for each chunk, writing to a separate file. It should be a quite simple add-on to the current patch, but will make life so much easier for parallel pg_dump. It would also be helpful to work around file size limitations on some filesystems. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On Thu, Dec 16, 2010 at 12:48 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: One more thing: the motivation behind this patch is to allow parallel pg_dump in the future, so we should be make sure this patch caters well for that. As soon as we have parallel pg_dump, the next big thing is going to be parallel dump of the same table using multiple processes. Perhaps we should prepare for that in the directory archive format, by allowing the data of a single table to be split into multiple files. That way parallel pg_dump is simple, you just split the table in chunks of roughly the same size, say 10GB each, and launch a process for each chunk, writing to a separate file. It should be a quite simple add-on to the current patch, but will make life so much easier for parallel pg_dump. It would also be helpful to work around file size limitations on some filesystems. Sounds reasonable. Are you planning to do this and commit? -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 16.12.2010 19:58, Robert Haas wrote: On Thu, Dec 16, 2010 at 12:48 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: One more thing: the motivation behind this patch is to allow parallel pg_dump in the future, so we should be make sure this patch caters well for that. As soon as we have parallel pg_dump, the next big thing is going to be parallel dump of the same table using multiple processes. Perhaps we should prepare for that in the directory archive format, by allowing the data of a single table to be split into multiple files. That way parallel pg_dump is simple, you just split the table in chunks of roughly the same size, say 10GB each, and launch a process for each chunk, writing to a separate file. It should be a quite simple add-on to the current patch, but will make life so much easier for parallel pg_dump. It would also be helpful to work around file size limitations on some filesystems. Sounds reasonable. Are you planning to do this and commit? I'll defer to Joachim, assuming he has the time energy. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On Thu, Dec 16, 2010 at 12:48 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: As soon as we have parallel pg_dump, the next big thing is going to be parallel dump of the same table using multiple processes. Perhaps we should prepare for that in the directory archive format, by allowing the data of a single table to be split into multiple files. That way parallel pg_dump is simple, you just split the table in chunks of roughly the same size, say 10GB each, and launch a process for each chunk, writing to a separate file. How exactly would you just split the table in chunks of roughly the same size ? Which queries should pg_dump send to the backend? If it just sends a bunch of WHERE queries, the server would still scan the same data several times since each pg_dump client would result in a seqscan over the full table. Ideally pg_dump should be able to query for all data in only one relation segment so that each segment is scanned by only one backend process. However this requires backend support and we would be sending queries that we'd not want clients other than pg_dump to send... If you were thinking about WHERE queries to get equally sized partitions, how would we deal with unindexed and/or non-numerical data in a large table? Joachim -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 16.12.2010 20:33, Joachim Wieland wrote: On Thu, Dec 16, 2010 at 12:48 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: As soon as we have parallel pg_dump, the next big thing is going to be parallel dump of the same table using multiple processes. Perhaps we should prepare for that in the directory archive format, by allowing the data of a single table to be split into multiple files. That way parallel pg_dump is simple, you just split the table in chunks of roughly the same size, say 10GB each, and launch a process for each chunk, writing to a separate file. How exactly would you just split the table in chunks of roughly the same size ? Check pg_class.relpages, and divide that evenly across the processes. That should be good enough. Which queries should pg_dump send to the backend? If it just sends a bunch of WHERE queries, the server would still scan the same data several times since each pg_dump client would result in a seqscan over the full table. Hmm, I was thinking of SELECT * FROM table WHERE ctid BETWEEN ? AND ?, but we don't support TidScans for ranges. Perhaps we could add that. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Heikki Linnakangas heikki.linnakan...@enterprisedb.com writes: On 16.12.2010 20:33, Joachim Wieland wrote: How exactly would you just split the table in chunks of roughly the same size ? Check pg_class.relpages, and divide that evenly across the processes. That should be good enough. Not even close ... relpages could be badly out of date. If you believe it, you could fail to dump data that's in further-out pages. We'd need to move pg_relpages() or some equivalent into core to make this workable. Which queries should pg_dump send to the backend? Hmm, I was thinking of SELECT * FROM table WHERE ctid BETWEEN ? AND ?, but we don't support TidScans for ranges. Perhaps we could add that. Yeah, that seems probably workable, given an up-to-date idea of the possible block range. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On Thu, Dec 16, 2010 at 2:29 PM, Tom Lane t...@sss.pgh.pa.us wrote: Heikki Linnakangas heikki.linnakan...@enterprisedb.com writes: On 16.12.2010 20:33, Joachim Wieland wrote: How exactly would you just split the table in chunks of roughly the same size ? Check pg_class.relpages, and divide that evenly across the processes. That should be good enough. Not even close ... relpages could be badly out of date. If you believe it, you could fail to dump data that's in further-out pages. We'd need to move pg_relpages() or some equivalent into core to make this workable. Which queries should pg_dump send to the backend? Hmm, I was thinking of SELECT * FROM table WHERE ctid BETWEEN ? AND ?, but we don't support TidScans for ranges. Perhaps we could add that. Yeah, that seems probably workable, given an up-to-date idea of the possible block range. So how bad would it be if we committed this new format without support for splitting large relations into multiple files, or with some stub support that never actually gets used, and fixed this later? Because this is starting to sound like a bigger project than I think we ought to be requiring for this patch. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 16.12.2010 22:13, Robert Haas wrote: So how bad would it be if we committed this new format without support for splitting large relations into multiple files, or with some stub support that never actually gets used, and fixed this later? Because this is starting to sound like a bigger project than I think we ought to be requiring for this patch. Would probably be fine, as long as we don't paint ourselves in the corner. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 12/16/2010 03:13 PM, Robert Haas wrote: So how bad would it be if we committed this new format without support for splitting large relations into multiple files, or with some stub support that never actually gets used, and fixed this later? Because this is starting to sound like a bigger project than I think we ought to be requiring for this patch. I don't think we have to have that in the first go at all. Parallel dump could be extremely useful without it. I haven't looked closely, but I assume there will still be an archive version recorded somewhere. When we change the archive format, bump the version number. cheers andrew -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Andrew Dunstan and...@dunslane.net writes: On 12/16/2010 03:13 PM, Robert Haas wrote: So how bad would it be if we committed this new format without support for splitting large relations into multiple files, or with some stub support that never actually gets used, and fixed this later? Because this is starting to sound like a bigger project than I think we ought to be requiring for this patch. I don't think we have to have that in the first go at all. Parallel dump could be extremely useful without it. I haven't looked closely, but I assume there will still be an archive version recorded somewhere. When we change the archive format, bump the version number. Sure, but it's worth thinking about the feature now. If there are format tweaks to be made, it might be less painful to make them now instead of later, even if actual support for the feature isn't there. (I agree I don't want to try to implement it just yet.) regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On Thursday 16 December 2010 19:33:10 Joachim Wieland wrote: On Thu, Dec 16, 2010 at 12:48 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: As soon as we have parallel pg_dump, the next big thing is going to be parallel dump of the same table using multiple processes. Perhaps we should prepare for that in the directory archive format, by allowing the data of a single table to be split into multiple files. That way parallel pg_dump is simple, you just split the table in chunks of roughly the same size, say 10GB each, and launch a process for each chunk, writing to a separate file. How exactly would you just split the table in chunks of roughly the same size ? Which queries should pg_dump send to the backend? If it just sends a bunch of WHERE queries, the server would still scan the same data several times since each pg_dump client would result in a seqscan over the full table. I would suggest implementing support for tidscans and doing it in segment size... Andres -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 17.12.2010 00:29, Andres Freund wrote: On Thursday 16 December 2010 19:33:10 Joachim Wieland wrote: On Thu, Dec 16, 2010 at 12:48 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: As soon as we have parallel pg_dump, the next big thing is going to be parallel dump of the same table using multiple processes. Perhaps we should prepare for that in the directory archive format, by allowing the data of a single table to be split into multiple files. That way parallel pg_dump is simple, you just split the table in chunks of roughly the same size, say 10GB each, and launch a process for each chunk, writing to a separate file. How exactly would you just split the table in chunks of roughly the same size ? Which queries should pg_dump send to the backend? If it just sends a bunch of WHERE queries, the server would still scan the same data several times since each pg_dump client would result in a seqscan over the full table. I would suggest implementingsupport for tidscans and doing it in segment size... I don't think there's any particular gain from matching the server's data file segment size, although 1GB does sound like a good chunk size for this too. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On Thursday 16 December 2010 23:34:02 Heikki Linnakangas wrote: On 17.12.2010 00:29, Andres Freund wrote: On Thursday 16 December 2010 19:33:10 Joachim Wieland wrote: On Thu, Dec 16, 2010 at 12:48 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: As soon as we have parallel pg_dump, the next big thing is going to be parallel dump of the same table using multiple processes. Perhaps we should prepare for that in the directory archive format, by allowing the data of a single table to be split into multiple files. That way parallel pg_dump is simple, you just split the table in chunks of roughly the same size, say 10GB each, and launch a process for each chunk, writing to a separate file. How exactly would you just split the table in chunks of roughly the same size ? Which queries should pg_dump send to the backend? If it just sends a bunch of WHERE queries, the server would still scan the same data several times since each pg_dump client would result in a seqscan over the full table. I would suggest implementingsupport for tidscans and doing it in segment size... I don't think there's any particular gain from matching the server's data file segment size, although 1GB does sound like a good chunk size for this too. Its noticeable more efficient reading from different files in different processes in comparison to all hammering the same file. Andres -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 12/16/2010 03:52 PM, Tom Lane wrote: Andrew Dunstanand...@dunslane.net writes: On 12/16/2010 03:13 PM, Robert Haas wrote: So how bad would it be if we committed this new format without support for splitting large relations into multiple files, or with some stub support that never actually gets used, and fixed this later? Because this is starting to sound like a bigger project than I think we ought to be requiring for this patch. I don't think we have to have that in the first go at all. Parallel dump could be extremely useful without it. I haven't looked closely, but I assume there will still be an archive version recorded somewhere. When we change the archive format, bump the version number. Sure, but it's worth thinking about the feature now. If there are format tweaks to be made, it might be less painful to make them now instead of later, even if actual support for the feature isn't there. (I agree I don't want to try to implement it just yet.) Yeah, OK. Well, time is getting short but (hand waving wildly) I think we could probably get by with just adding a member to the TOC for the section number of the entry (set it to 0 for non TABLE DATA TOC entries). The section number could be built into the file name in directory format. For now that number would always be 1 for TABLE DATA members. This has intriguing possibilities for parallel restore of custom format dumps too. It could be very useful to be able to restore a single table in parallel, if we had more than one TABLE DATA member per table. I'm deliberately just addressing infrastructure issues rather than how we actually generate multiple sections of data for a single table (especially if we want to do that in parallel). cheers andrew -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On Thu, Dec 2, 2010 at 2:52 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: Ok, committed, with some small cleanup since the last patch I posted. Could you update the directory-format patch on top of the committed version, please? Thanks for committing the first part. Here is the updated and rebased directory-format patch. Joachim pg_dump-directory-rebased.diff.gz Description: GNU Zip compressed data -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 02.12.2010 23:12, Alvaro Herrera wrote: Excerpts from Heikki Linnakangas's message of jue dic 02 16:52:27 -0300 2010: Ok, committed, with some small cleanup since the last patch I posted. I think the comments on _ReadBuf and friends need to be updated, since they are not just for headers and TOC stuff anymore. I'm not sure if they were already outdated before your patch ... These routines are only used to read write headers TOC Hmm, ReadInt calls _ReadByte, and PrintData used to call ReadInt, so it was indirectly been called for things other than headers and TOC already. Unless you consider the headers to include length integer in in each data block. I'm inclined to just remove that sentence. I also note that the _Clone and _DeClone functions are a bit misplaced. There's a big END OF FORMAT CALLBACKS earlier in the file, but _Clone and _DeClone are such callbacks. I'll move them to the right place. PS. Thanks for the cleanup you did yesterday. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Ok, committed, with some small cleanup since the last patch I posted. Could you update the directory-format patch on top of the committed version, please? -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Excerpts from Heikki Linnakangas's message of jue dic 02 16:52:27 -0300 2010: Ok, committed, with some small cleanup since the last patch I posted. I think the comments on _ReadBuf and friends need to be updated, since they are not just for headers and TOC stuff anymore. I'm not sure if they were already outdated before your patch ... -- Álvaro Herrera alvhe...@commandprompt.com The PostgreSQL Company - Command Prompt, Inc. PostgreSQL Replication, Consulting, Custom Development, 24x7 support -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 29.11.2010 22:21, Heikki Linnakangas wrote: On 29.11.2010 07:11, Joachim Wieland wrote: On Mon, Nov 22, 2010 at 3:44 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: * wrap long lines * use extern in function prototypes in header files * inline some functions like _StartDataCompressor, _EndDataCompressor, _DoInflate/_DoDeflate that aren't doing anything but call some other function. So here is a new round of patches. It turned out that the feature to allow to also restore files from a different dump and with a different compression required some changes in the compressor API. And in the end I didn't like all the #ifdefs either and made a less #ifdef-rich version using function pointers. Ok. The separate InitCompressorState() and AllocateCompressorState() functions seem unnecessary. As the code stands, there's little performance gain from re-using the same CompressorState, just re-initializing it, and I can't see any other justification for them either. I combined those, and the Free/Flush steps, and did a bunch of other editorializations and cleanups. Here's an updated patch, also available in my git repository at git://git.postgresql.org/git/users/heikki/postgres.git, branch pg_dump-dir. I'm going to continue reviewing this later, tomorrow hopefully. Here's another update. I changed things quite heavily. I didn't see the point of having the Alloc+Free functions for uncompressing, because the ReadDataFromArchive processed the whole input stream in one go anyway. So the new API consists of four functions, AllocateCompressor, WriteDataToArchive and EndCompressor for writing, and ReadDataFromArchive for reading. Also, I reverted the zlib buffer size from 64k to 4k. If you want to raise that, let's discuss that separately. Please let me know what you think of this version, or if you spot any bugs. I'll keep working on this, I'm hoping to get this into committable shape by the end of the week. The pg_backup_directory patch naturally won't apply over this anymore. Once we have the compress_io part in shape, that will need to be fixed. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 01.12.2010 16:03, Heikki Linnakangas wrote: On 29.11.2010 22:21, Heikki Linnakangas wrote: I combined those, and the Free/Flush steps, and did a bunch of other editorializations and cleanups. Here's an updated patch, also available in my git repository at git://git.postgresql.org/git/users/heikki/postgres.git, branch pg_dump-dir. I'm going to continue reviewing this later, tomorrow hopefully. Here's another update. Forgot attachment. This is also available in the above git repo. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com *** a/src/bin/pg_dump/Makefile --- b/src/bin/pg_dump/Makefile *** *** 20,26 override CPPFLAGS := -I$(libpq_srcdir) $(CPPFLAGS) OBJS= pg_backup_archiver.o pg_backup_db.o pg_backup_custom.o \ pg_backup_files.o pg_backup_null.o pg_backup_tar.o \ ! dumputils.o $(WIN32RES) KEYWRDOBJS = keywords.o kwlookup.o --- 20,26 OBJS= pg_backup_archiver.o pg_backup_db.o pg_backup_custom.o \ pg_backup_files.o pg_backup_null.o pg_backup_tar.o \ ! dumputils.o compress_io.o $(WIN32RES) KEYWRDOBJS = keywords.o kwlookup.o *** /dev/null --- b/src/bin/pg_dump/compress_io.c *** *** 0 --- 1,404 + /*- + * + * compress_io.c + * Routines for archivers to write an uncompressed or compressed data + * stream. + * + * Portions Copyright (c) 1996-2010, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * The interface for writing to an archive consists of three functions: + * AllocateCompressor, WriteDataToArchive and EndCompressor. First you call + * AllocateCompressor, then write all the data by calling WriteDataToArchive + * as many times as needed, and finally EndCompressor. WriteDataToArchive + * and EndCompressor will call the WriteFunc that was provided to + * AllocateCompressor for each chunk of compressed data. + * + * The interface for reading an archive consists of just one function: + * ReadDataFromArchive. ReadDataFromArchive reads the whole compressed input + * stream, by repeatedly calling the given ReadFunc. ReadFunc returns the + * compressed data chunk at a time, and ReadDataFromArchive decompresses it + * and passes the decompressed data to ahwrite(), until ReadFunc returns 0 + * to signal EOF. + * + * The interface is the same for compressed and uncompressed streams. + * + * + * IDENTIFICATION + * src/bin/pg_dump/compress_io.c + * + *- + */ + + #include compress_io.h + + static const char *modulename = gettext_noop(compress_io); + + static void ParseCompressionOption(int compression, CompressorAlgorithm *alg, + int *level); + + /* Routines that are private to a specific compressor (static functions) */ + #ifdef HAVE_LIBZ + /* Routines that support zlib compressed data I/O */ + static void InitCompressorZlib(CompressorState *cs, int level); + static void DeflateCompressorZlib(ArchiveHandle *AH, CompressorState *cs, + bool flush); + static void ReadDataFromArchiveZlib(ArchiveHandle *AH, ReadFunc readF); + static size_t WriteDataToArchiveZlib(ArchiveHandle *AH, CompressorState *cs, + const char *data, size_t dLen); + static void EndCompressorZlib(ArchiveHandle *AH, CompressorState *cs); + + #endif + + /* Routines that support uncompressed data I/O */ + static void ReadDataFromArchiveNone(ArchiveHandle *AH, ReadFunc readF); + static size_t WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, + const char *data, size_t dLen); + + static void + ParseCompressionOption(int compression, CompressorAlgorithm *alg, int *level) + { + /* + * The compression is set either on the commandline when creating + * an archive or by ReadHead() when restoring an archive. It can also be + * set on a per-data item basis in the directory archive format. + */ + if (compression == Z_DEFAULT_COMPRESSION || + (compression 0 compression = 9)) + *alg = COMPR_ALG_LIBZ; + else if (compression == 0) + *alg = COMPR_ALG_NONE; + else + die_horribly(NULL, modulename, Invalid compression code: %d\n, + compression); + + if (level) + *level = compression; + } + + /* Public interface routines */ + + /* Allocate a new compressor */ + CompressorState * + AllocateCompressor(int compression, WriteFunc writeF) + { + CompressorState *cs; + CompressorAlgorithm alg; + int level; + + ParseCompressionOption(compression, alg, level); + + cs = (CompressorState *) calloc(1, sizeof(CompressorState)); + if (cs == NULL) + die_horribly(NULL, modulename, out of memory\n); + cs-writeF = writeF; + cs-comprAlg = alg; + + #ifndef HAVE_LIBZ + if (alg == COMPR_ALG_LIBZ) + die_horribly(NULL, modulename, not built with zlib support\n); + #endif + + /* + * Perform compression algorithm specific
Re: [HACKERS] directory archive format for pg_dump
On Wed, Dec 1, 2010 at 9:05 AM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: Forgot attachment. This is also available in the above git repo. I have quickly checked your modifications, on the one hand I like the reduction of functions, I would have said that we have AH around all the time and so we could just allocate once and stuff it all into ctx-cs and reuse the buffers for every object, but re-allocating them for every (dumpable) object should be fine as well. Regarding the function pointers that you removed, you are now putting back in what Dimitri wanted me to take out, namely switch/case instructions for the algorithms and then #ifdefs for every algorithm. It's not too many now since we have taken out LZF. Well, I can live with both ways. There is one thing however that I am not in favor of, which is the removal of the sizeHint parameter for the read functions. The reason for this parameter is not very clear now without LZF but I have tried to put in a few comments to explain the situation (which you have taken out as well :-) ). The point is that zlib is a stream based compression algorithm, you just stuff data in and from time to time you get data out and in the end you explicitly flush the compressor. The read function can just return as many bytes as it wants and we can just hand it all over to zlib. Other compression algorithms however are block based and first write a block header that contains the information on the next data block, including uncompressed and compressed sizes. Now with the sizeHint parameter I used, the compressor could tell the read function that it just wants to read the fixed size header (6 bytes IIRC). In the header it would look up the compressed size for the next block and would then ask the read function to get exactly this amount of data, decompress it and go on with the next block, and so forth... Of course you can possibly do that memory management inside the compressor with an extra buffer holding what you got in excess but it's a pain. If you removed that part on purpose on the grounds that there is no block based compression algorithm in core and probably never will be, then that's okay :-) Joachim -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 02.12.2010 04:35, Joachim Wieland wrote: There is one thing however that I am not in favor of, which is the removal of the sizeHint parameter for the read functions. The reason for this parameter is not very clear now without LZF but I have tried to put in a few comments to explain the situation (which you have taken out as well :-) ). The point is that zlib is a stream based compression algorithm, you just stuff data in and from time to time you get data out and in the end you explicitly flush the compressor. The read function can just return as many bytes as it wants and we can just hand it all over to zlib. Other compression algorithms however are block based and first write a block header that contains the information on the next data block, including uncompressed and compressed sizes. Now with the sizeHint parameter I used, the compressor could tell the read function that it just wants to read the fixed size header (6 bytes IIRC). In the header it would look up the compressed size for the next block and would then ask the read function to get exactly this amount of data, decompress it and go on with the next block, and so forth... Of course you can possibly do that memory management inside the compressor with an extra buffer holding what you got in excess but it's a pain. If you removed that part on purpose on the grounds that there is no block based compression algorithm in core and probably never will be, then that's okay :-) Yeah, we're not going to have lzf built-in anytime soon. The external command approach seems like the best way to support additional compression algorithms, and I don't think it could do anything with sizeHint. And the custom format didn't obey sizeHint anyway, because it reads one custom-format block at a time. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 29.11.2010 07:11, Joachim Wieland wrote: On Mon, Nov 22, 2010 at 3:44 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: * wrap long lines * use extern in function prototypes in header files * inline some functions like _StartDataCompressor, _EndDataCompressor, _DoInflate/_DoDeflate that aren't doing anything but call some other function. So here is a new round of patches. It turned out that the feature to allow to also restore files from a different dump and with a different compression required some changes in the compressor API. And in the end I didn't like all the #ifdefs either and made a less #ifdef-rich version using function pointers. The downside now is that I have created quite a few one-line functions that Heikki doesn't like all that much, but I assume that they are okay in this case on the grounds that the public compressor interface is calling the private implementation of a certain compressor. Thanks, I'll take a look. BTW, I know you wanted to have support for other compression algorithms; I think the best way to achieve that is to make it possible to specify an external command to be used for compression. pg_dump would fork() and exec() that, and pipe the data to be compressed/decompressed to stdin/stdout of the external command. We're not going to add support for every new compression algorithm that's in vogue, but generic external command support should make happy those who want it. I'd be particularly excited about using something like pbzip2, to speed up the compression on multi-core systems. That should be a separate patch, but it's something to keep in mind with these refactorings. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On Mon, Nov 29, 2010 at 10:49 AM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: On 29.11.2010 07:11, Joachim Wieland wrote: On Mon, Nov 22, 2010 at 3:44 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: * wrap long lines * use extern in function prototypes in header files * inline some functions like _StartDataCompressor, _EndDataCompressor, _DoInflate/_DoDeflate that aren't doing anything but call some other function. So here is a new round of patches. It turned out that the feature to allow to also restore files from a different dump and with a different compression required some changes in the compressor API. And in the end I didn't like all the #ifdefs either and made a less #ifdef-rich version using function pointers. The downside now is that I have created quite a few one-line functions that Heikki doesn't like all that much, but I assume that they are okay in this case on the grounds that the public compressor interface is calling the private implementation of a certain compressor. Thanks, I'll take a look. BTW, I know you wanted to have support for other compression algorithms; I think the best way to achieve that is to make it possible to specify an external command to be used for compression. pg_dump would fork() and exec() that, and pipe the data to be compressed/decompressed to stdin/stdout of the external command. We're not going to add support for every new compression algorithm that's in vogue, but generic external command support should make happy those who want it. I'd be particularly excited about using something like pbzip2, to speed up the compression on multi-core systems. That should be a separate patch, but it's something to keep in mind with these refactorings. That would also ease licensing concerns, since we wouldn't have to redistribute or bundle anything. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 29.11.2010 07:11, Joachim Wieland wrote: On Mon, Nov 22, 2010 at 3:44 PM, Heikki Linnakangas heikki.linnakan...@enterprisedb.com wrote: * wrap long lines * use extern in function prototypes in header files * inline some functions like _StartDataCompressor, _EndDataCompressor, _DoInflate/_DoDeflate that aren't doing anything but call some other function. So here is a new round of patches. It turned out that the feature to allow to also restore files from a different dump and with a different compression required some changes in the compressor API. And in the end I didn't like all the #ifdefs either and made a less #ifdef-rich version using function pointers. Ok. The separate InitCompressorState() and AllocateCompressorState() functions seem unnecessary. As the code stands, there's little performance gain from re-using the same CompressorState, just re-initializing it, and I can't see any other justification for them either. I combined those, and the Free/Flush steps, and did a bunch of other editorializations and cleanups. Here's an updated patch, also available in my git repository at git://git.postgresql.org/git/users/heikki/postgres.git, branch pg_dump-dir. I'm going to continue reviewing this later, tomorrow hopefully. The downside now is that I have created quite a few one-line functions that Heikki doesn't like all that much, but I assume that they are okay in this case on the grounds that the public compressor interface is calling the private implementation of a certain compressor. You could avoid the wrapper functions by calling the function pointers directly, but I agree it seems neater the way you did it. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com *** a/src/bin/pg_dump/Makefile --- b/src/bin/pg_dump/Makefile *** *** 20,26 override CPPFLAGS := -I$(libpq_srcdir) $(CPPFLAGS) OBJS= pg_backup_archiver.o pg_backup_db.o pg_backup_custom.o \ pg_backup_files.o pg_backup_null.o pg_backup_tar.o \ ! dumputils.o $(WIN32RES) KEYWRDOBJS = keywords.o kwlookup.o --- 20,26 OBJS= pg_backup_archiver.o pg_backup_db.o pg_backup_custom.o \ pg_backup_files.o pg_backup_null.o pg_backup_tar.o \ ! dumputils.o compress_io.o $(WIN32RES) KEYWRDOBJS = keywords.o kwlookup.o *** /dev/null --- b/src/bin/pg_dump/compress_io.c *** *** 0 --- 1,415 + /*- + * + * compress_io.c + * Routines for archivers to write an uncompressed or compressed data + * stream. + * + * Portions Copyright (c) 1996-2010, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_io.c + * + *- + */ + + #include compress_io.h + + static const char *modulename = gettext_noop(compress_io); + + /* Routines that are private to a specific compressor (static functions) */ + #ifdef HAVE_LIBZ + /* Routines that support zlib compressed data I/O */ + static void InitCompressorZlib(CompressorState *cs, int compression); + static void DeflateCompressorZlib(ArchiveHandle *AH, CompressorState *cs, + bool flush); + static void ReadDataFromArchiveZlib(ArchiveHandle *AH, CompressorState *cs); + static size_t WriteDataToArchiveZlib(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen); + static void EndCompressorZlib(ArchiveHandle *AH, CompressorState *cs); + static CompressorState *AllocateCompressorState(CompressorAction action, + int compression); + + static CompressorFuncs cfs_zlib = { + InitCompressorZlib, + ReadDataFromArchiveZlib, + WriteDataToArchiveZlib, + EndCompressorZlib + }; + #endif + + /* Routines that support uncompressed data I/O */ + static void InitCompressorNone(CompressorState *cs, int compression); + static void ReadDataFromArchiveNone(ArchiveHandle *AH, CompressorState *cs); + static size_t WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen); + static void EndCompressorNone(ArchiveHandle *AH, CompressorState *cs); + + static CompressorFuncs cfs_none = { + InitCompressorNone, + ReadDataFromArchiveNone, + WriteDataToArchiveNone, + EndCompressorNone + }; + + /* Allocate a new decompressor */ + CompressorState * + AllocateInflator(int compression, ReadFunc readF) + { + CompressorState *cs; + + cs = AllocateCompressorState(COMPRESSOR_INFLATE, compression); + cs-readF = readF; + + return cs; + } + + /* Allocate a new compressor */ + CompressorState * + AllocateDeflator(int compression, WriteFunc writeF) + { + CompressorState *cs; + + cs = AllocateCompressorState(COMPRESSOR_DEFLATE, compression); + cs-writeF = writeF; + + return cs; + } + + static CompressorState * + AllocateCompressorState(CompressorAction action, int compression) + {
Re: [HACKERS] directory archive format for pg_dump
On 20.11.2010 06:10, Joachim Wieland wrote: 2010/11/19 José Arthur Benetasso Villanovajose.art...@gmail.com: The md5.c and kwlookup.c reuse using a link doesn't look nice either. This way you need to compile twice, among others things, but I think that its temporary, right? No, it isn't. md5.c is used in the same way by e.g. libpq and there are other examples for links in core, check out src/bin/psql for example. It seems like overkill to include md5 just for hashing the random bytes that getRandomData() generates. And if random() doesn't produce unique values, it's not going to get better by hashing it. How about using a timestamp instead of the hash? If you don't initialize random() with srandom(), BTW, it will always return the same value. But I'm not actually sure we should be preventing mix match of files from different dumps. It might be very useful to do just that sometimes, like restoring a recent backup, with the contents of one table replaced with older data. A warning would be ok, though. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Heikki Linnakangas heikki.linnakan...@enterprisedb.com writes: But I'm not actually sure we should be preventing mix match of files from different dumps. It might be very useful to do just that sometimes, like restoring a recent backup, with the contents of one table replaced with older data. A warning would be ok, though. +1. This mechanism seems like a solution in search of a problem. Just lose the whole thing, and instead fix pg_dump to complain if the target directory isn't empty. That should be sufficient to guard against accidental mixing of different dumps, and as Heikki says there's not a good reason to prevent intentional mixing. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On 22.11.2010 19:07, Tom Lane wrote: Heikki Linnakangasheikki.linnakan...@enterprisedb.com writes: But I'm not actually sure we should be preventing mix match of files from different dumps. It might be very useful to do just that sometimes, like restoring a recent backup, with the contents of one table replaced with older data. A warning would be ok, though. +1. This mechanism seems like a solution in search of a problem. Just lose the whole thing, and instead fix pg_dump to complain if the target directory isn't empty. That should be sufficient to guard against accidental mixing of different dumps, and as Heikki says there's not a good reason to prevent intentional mixing. Extending that thought a bit, it would be nice if the per-file header would carry the info if the file is compressed or not, instead of just one such flag in the TOC. You could then also mix match files from compressed and non-compressed archives. Other than the md5 thing, the patch looks fine to me. There's many quite levels of indirection, it took me a while to get my head around the call chains like DataDumper-_WriteData-WriteDataToArchive-_WriteBuf, but I don't have any ideas on how to improve that. However, docs are missing, so I'm marking this as Waiting on Author. There's some cosmetic changes I'd like to have fixed or do myself before committing: * wrap long lines * use extern in function prototypes in header files * inline some functions like _StartDataCompressor, _EndDataCompressor, _DoInflate/_DoDeflate that aren't doing anything but call some other function. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Hi, Sharing some thoughts after a first round of reviewing, where I only had time to read the patch itself. Joachim Wieland j...@mcknight.de writes: Since the compression is currently all down in the custom format backup code, the first thing I've done was refactoring the compression functions into a separate file. While at it, I have added support for liblzf compression. I think I'd like to see a separate patch for the new compression support. Sorry about that, I realize that's extra work… And it could be about personal preferences, but the way you added the liblzf support strikes me at odd, with all those #ifdefs everywhere. Is it possible to have a specific file for each supported compression format, then some routing code in src/bin/pg_dump/compress_io.c? The routing code already exists but then the file is full of #ifdef sections to define the right supporting function when I think having a compress_io_zlib and a compress_io_lzf files would be better. Then there's the bulk of the new dump format feature in the other part of the patch, namely src/bin/pg_dump/pg_backup_directory.c. You have to update the copyright in the file header there, at least :) I'm yet to devote more time on this part of the patch but it seems like it's rewriting the full support without using the existing bits. That's something I have to check, didn't have time to read the existing other archive formats code there. I'm hesitant as far as marking the patch Waiting on author to get it split. Joachim, what do you think? Regards, -- Dimitri Fontaine http://2ndQuadrant.fr PostgreSQL : Expertise, Formation et Support -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Hi Dimitri and Joachim. I've looked the patch too, and I want to share some thoughts too. I've used http://wiki.postgresql.org/wiki/Reviewing_a_Patch to guide my review. Submission review: I've apllied and compiled the patch successfully using the current master. Usability review: The dir format generated in my database 60 files, with different sizes, and it looks very confusing. Is it possible to use the same trick as pigz and pbzip2, creating a concatenated file of streams? Feature test: Just a partial review. I can dump / restore using lzf, but didnt stress it hard to check robustness. Performance review: Didnt test it hard too, but looks ok. Coding review: Just a shallow review here. I think I'd like to see a separate patch for the new compression support. Sorry about that, I realize that's extra work… Same feeling here, this is the 1st thing that I notice. The md5.c and kwlookup.c reuse using a link doesn't look nice either. This way you need to compile twice, among others things, but I think that its temporary, right? -- José Arthur Benetasso Villanova -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Excerpts from José Arthur Benetasso Villanova's message of vie nov 19 18:28:03 -0300 2010: The md5.c and kwlookup.c reuse using a link doesn't look nice either. This way you need to compile twice, among others things, but I think that its temporary, right? Not sure what you mean here, but kwlookup.c is a symlink without this patch too. It's just the way it works; the compilation environments here and in the backend are different, so there is no other option but to compile twice. I guess md5.c is a new one (I didn't check), but I would assume it's the same thing. -- Álvaro Herrera alvhe...@commandprompt.com The PostgreSQL Company - Command Prompt, Inc. PostgreSQL Replication, Consulting, Custom Development, 24x7 support -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Hi Dimitri, thanks for reviewing my patch! On Fri, Nov 19, 2010 at 2:44 PM, Dimitri Fontaine dimi...@2ndquadrant.fr wrote: I think I'd like to see a separate patch for the new compression support. Sorry about that, I realize that's extra work… I guess it wouldn't be a very big deal but I also doubt that it makes the review that much easier. Basically the compression refactor patch would just touch pg_backup_custom.c (because this is the place where the libz compression is currently burried into) and the two new compress_io.(c|h) files. Everything else is pretty much the directory stuff and is on top of these changes. And it could be about personal preferences, but the way you added the liblzf support strikes me at odd, with all those #ifdefs everywhere. Is it possible to have a specific file for each supported compression format, then some routing code in src/bin/pg_dump/compress_io.c? Sure we could. But I wanted to wait with any fancy function pointer stuff until we have decided if we want to include the liblzf support at all. The #ifdefs might be a bit ugly but in case we do not include liblzf support, it's the easiest way to take it out again. As written in my introduction, this patch is not really about liblzf, liblzf is just a proof of concept for factoring out the compression part and I have included it, so that people can use it and see how much speed improvement they get. The routing code already exists but then the file is full of #ifdef sections to define the right supporting function when I think having a compress_io_zlib and a compress_io_lzf files would be better. Sure! I completely agree... Then there's the bulk of the new dump format feature in the other part of the patch, namely src/bin/pg_dump/pg_backup_directory.c. You have to update the copyright in the file header there, at least :) Well, not sure if we can just change the copyright notice, because in the end the structure was copied from one of the other files which all have the copyright notice in them, so my work is based on those other files... I'm hesitant as far as marking the patch Waiting on author to get it split. Joachim, what do you think? I will see if I can split it. Joachim -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
Dimitri Fontaine dimi...@2ndquadrant.fr writes: I think I'd like to see a separate patch for the new compression support. Sorry about that, I realize that's extra work⦠That part of the patch is likely to get rejected outright anyway, so I *strongly* recommend splitting it out. We have generally resisted adding random compression algorithms to pg_dump because of license and patent considerations, and I see no reason to suppose this one is going to pass muster. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] directory archive format for pg_dump
On Fri, Nov 19, 2010 at 11:53 PM, Tom Lane t...@sss.pgh.pa.us wrote: Dimitri Fontaine dimi...@2ndquadrant.fr writes: I think I'd like to see a separate patch for the new compression support. Sorry about that, I realize that's extra work… That part of the patch is likely to get rejected outright anyway, so I *strongly* recommend splitting it out. We have generally resisted adding random compression algorithms to pg_dump because of license and patent considerations, and I see no reason to suppose this one is going to pass muster. I was already anticipating that possiblitiy and my inital patch description is along these lines. However, liblzf is BSD licensed so on the license side we should be fine. Regarding patents, your last comment was that you'd like to see if it's really worth it and so I have included support for lzf for anybody to go ahead and find that out. Will send an updated split up patch this weekend (which would actually be four patches already...). Joachim
Re: [HACKERS] directory archive format for pg_dump
Hi Jose, 2010/11/19 José Arthur Benetasso Villanova jose.art...@gmail.com: The dir format generated in my database 60 files, with different sizes, and it looks very confusing. Is it possible to use the same trick as pigz and pbzip2, creating a concatenated file of streams? What pigz is parallelizing is the actual computation of the compressed data. The directory archive format however is a preparation for a parallel pg_dump, dumping several tables (especially large tables of course) in parallel via multiple database connections and multiple pg_dump frontends. The idea of multiplexing their output into one file has been rejected on the grounds that it would probably slow down the whole process. Nevertheless pigz could be implemented as an alternative compression algorithm and that way the custom and the directory archive format could use it, but here as well, license and patent questions might be in the way, even though it is based on libz. The md5.c and kwlookup.c reuse using a link doesn't look nice either. This way you need to compile twice, among others things, but I think that its temporary, right? No, it isn't. md5.c is used in the same way by e.g. libpq and there are other examples for links in core, check out src/bin/psql for example. Joachim -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers