This is the main PostgreSQL git repository. http://git.postgresql.org/cgit/postgresql.git/atom?h=REL8_2_STABLE 2015-01-20T15:26:20Z 2015-01-20T15:26:20Z Alvaro Herrera 2015-01-20T15:26:20Z urn:sha1:105f3ef492abd22985321b64e8f40e58f3a15a83 With this optimization flag enabled, recent versions of gcc can generate incorrect code that assumes variable-length arrays (such as oidvector) are actually fixed-length because they're embedded in some larger struct. The known instance of this problem was fixed in 9.2 and up by commit 8137f2c32322c624e0431fac1621e8e9315202f9 and followon work, which hides actually-variable-length catalog fields from the compiler altogether. And we plan to gradually convert variable-length fields to official "flexible array member" notation over time, which should prevent this type of bug from reappearing as gcc gets smarter. We're not going to try to back-port those changes into older branches, though, so apply this band-aid instead. Andres Freund This is a backpatch of commit 649839dd9 to unsupported branches REL8_2_STABLE and REL8_3_STABLE, so that they work with newer toolsets. 2011-12-14T22:05:12Z Andrew Dunstan 2011-12-14T22:05:12Z urn:sha1:91a1e1489986533aee663de41382f9ba8f8e39d9 Suggested solution from Tom Lane. Problem discovered, probably not for the first time, while testing the mingw-w64 32 bit compiler. Backpatched to all live branches. 2011-12-01T21:59:19Z Tom Lane 2011-12-01T21:59:19Z urn:sha1:fa1369a6b901bc1f7efa3d8386ffe5102f78c78a Hail and farewell, 8.2. 2011-09-22T22:09:27Z Tom Lane 2011-09-22T22:09:27Z urn:sha1:d08aa1937fbdf756703bbb2653538b06541ba349 2011-08-29T23:52:13Z Tom Lane 2011-08-29T23:52:13Z urn:sha1:cbd51548dc1869fde58b621ea3c2fc9386947ea9 The version of this macro used in autoconf 2.59 is capable of incorrectly succeeding (ie, reporting that a library function is available when it isn't), if the compiler performs link-time optimization and decides that it can optimize the function reference away entirely. Replace it with the coding used in autoconf 2.61 and later, which forces the program result to depend on the function's result so that it cannot be optimized away. This should fix build failures currently being seen on buildfarm member anchovy. This patch affects the 8.2 and 8.3 branches only, since later branches are using autoconf versions that don't have this problem. 2011-06-14T21:14:11Z Tom Lane 2011-06-14T21:14:11Z urn:sha1:07bc6fe6687fbf2ac0fea5b09fe5a4932a725368 We previously found out that OS X's standard perl installation tries to put -arch switches into Perl link commands, evidently in hopes of building universal binaries. But it doesn't work to add such switches in plperl's link step if they weren't being used earlier, so this is basically unworkable. When using gcc the result is only some warnings; but LLVM fails entirely, so this issue isn't as cosmetic as we originally thought. Hence, back-patch commit d69a419e682c2d39c2355105a7e5e2b90357c8f0 into pre-9.0 branches. 2011-04-15T03:19:01Z Marc G. Fournier 2011-04-15T03:19:01Z urn:sha1:1e289824b6fa9a2ce7ed13d8559f05683a75f466 2011-01-28T02:25:42Z Marc G. Fournier 2011-01-28T02:25:42Z urn:sha1:f62f223b8e7519d741203d71f5bd8fc7693c0b41 2010-12-16T04:51:07Z Tom Lane 2010-12-16T04:51:07Z urn:sha1:6a61d58ca8d28b1d0e49a6906ac61ab86f12493c The mingw people don't appear to care about compatibility with non-GNU versions of getopt, so force use of our own copy of getopt on Windows. Also, ensure that we make use of optreset when using our own copy. Per report from Andrew Dunstan. Back-patch to all versions supported on Windows. 2010-12-14T03:03:33Z Marc G. Fournier 2010-12-14T03:03:33Z urn:sha1:4d836f34b51e3726c8765aab1feff98a185e6a5d