boost/build
2
0
Fork 0
mirror of https://github.com/boostorg/build.git synced 2026-02-16 13:22:11 +00:00
Code Issues Projects Releases Wiki Activity
4,333 Commits 2 Branches 134 Tags
fc6a60ee1e4ec0d4a93aa9ff7d712d1b51396bba
Commit Graph

730 Commits

Author SHA1 Message Date
Jurko Gospodnetić
fc6a60ee1e Minor stylistic changes. Removed some minor code duplication. 
[SVN r78862]
 2012-06-09 07:28:42 +00:00
K. Noel Belcourt
c851e03de3 Handle signals causing early returns from select 
and waitpid.  Should fix FreeBSD reported problem.



[SVN r78861]
 2012-06-09 00:21:03 +00:00
K. Noel Belcourt
a8344266f3 Add ability to limit amount of target output that is 
output by bjam.  Added bjam argument -mx where x is
the maximum amount of output to be captured from a 
target, in kb.

This fix will enable, PGI, VACPP, and Clang to cycle
normally in the nightly testing.



[SVN r78847]
 2012-06-07 15:36:46 +00:00
K. Noel Belcourt
6e89b4abac Revert commit 76862, restore FreeBSD and other 
unix testers that I broke with this commit.



[SVN r78800]
 2012-06-02 18:26:35 +00:00
Jurko Gospodnetić
564fda1ae6 Updated the SPLIT_BY_CHARACTERS builtin rule signature so invalid calls get rejected with a correct error message instead of silently causing the rule to return an empty list. 
[SVN r78608]
 2012-05-25 10:16:36 +00:00
Jurko Gospodnetić
0c65651300 Minor stylistic change - trimmed trailing whitespace. 
[SVN r78576]
 2012-05-24 09:56:23 +00:00
Jurko Gospodnetić
0b1a787f20 Fixed Boost Jam crash when the SPLIT_BY_CHARACTERS rule gets passed an empty list as any of its arguments. 
[SVN r78575]
 2012-05-24 09:53:24 +00:00
Jurko Gospodnetić
10bf219ca4 Explicitly flushed bjam's standard output after printing out '...patience...' lines. Without a fflush() call these lines may get buffered and displayed all at once only after the first actual target has been built. It seems that Visual C++ compiler's runtime does not use line buffering for its standard output stream if given a pipe as its standard output (encountered with bjam compiled using msvc 10 and run by piping its output through the Cygwin tee utility). 
[SVN r78545]
 2012-05-22 21:06:47 +00:00
Steven Watanabe
c770907c69 Avoid a spurious dependency. 
[SVN r78302]
 2012-05-02 06:55:44 +00:00
Steven Watanabe
64add142e1 Replace ad hoc (incorrect) cycle detection code with a variation of Tarjan's algorithm. 
[SVN r78287]
 2012-05-01 06:45:35 +00:00
Steven Watanabe
e7e546092b Avoid a self-cycle in scc_root. 
[SVN r78283]
 2012-04-30 18:21:47 +00:00
Steven Watanabe
2ddd4f5f69 Fix an infinite loop caused by creating a cycle in the strongly connected component links. 
[SVN r78282]
 2012-04-30 18:11:35 +00:00
Steven Watanabe
656fffe626 Don't assume that make1a processes targets in the same order as make0. It doesn't. Start a new stack when rescanning to avoid spurious cycles which cause a seg-fault. 
[SVN r78281]
 2012-04-30 16:49:42 +00:00
Steven Watanabe
d3b0ec3508 Handle cycles when determining target fate in make0. 
[SVN r78259]
 2012-04-30 01:21:04 +00:00
Steven Watanabe
4fbd8fa775 Pass the correct argument to make1breakcycle and make sure that we don't add a spurious self-dependency. 
[SVN r78252]
 2012-04-29 04:27:55 +00:00
Steven Watanabe
9e04ceafc3 Stop targets that depend on an include cycle from updating before all the required generated headers are built. 
[SVN r78250]
 2012-04-29 01:06:55 +00:00
Steven Watanabe
e6b3927ef3 Detect cycles created by rescanning. 
[SVN r78249]
 2012-04-28 22:05:28 +00:00
Steven Watanabe
c4465a1366 Use a regular dependency instead of an include dependency for generated headers to allow correct rescanning of generated headers. 
[SVN r78247]
 2012-04-28 19:23:47 +00:00
Steven Watanabe
a72ec8b40a Don't rescan targets when their internal include node is updated. 
[SVN r78246]
 2012-04-28 18:19:12 +00:00
Steven Watanabe
ca19ccd738 make sure that we don't call make0 twice on the original include node. 
[SVN r78244]
 2012-04-28 17:39:06 +00:00
Steven Watanabe
32640ce526 Be sure to use object_equal. 
[SVN r77727]
 2012-04-02 22:27:28 +00:00
Steven Watanabe
9345ea38fa Fix cleanup in hcache_init. 
[SVN r77726]
 2012-04-02 22:24:14 +00:00
Steven Watanabe
7fca6a5c1d Fix memory errors with BJAM_NO_MEM_CACHE. 
[SVN r77725]
 2012-04-02 22:00:32 +00:00
Steven Watanabe
9501c61e8c Combine -d+10 stats for INSTANCE modules by class. Make stats correct for a separate chaining hash table. The old version seems to have been for open addressing. 
[SVN r77657]
 2012-03-30 21:34:39 +00:00
Steven Watanabe
a686b2082f Remove unused variable. 
[SVN r77654]
 2012-03-30 19:12:25 +00:00
Steven Watanabe
4ef470dbf5 Remove unused variable. 
[SVN r77647]
 2012-03-30 14:05:22 +00:00
Steven Watanabe
771a58f19a Fix seg-fault. 
[SVN r77645]
 2012-03-30 13:49:00 +00:00
Steven Watanabe
c42418c528 Workoround for xlc. Patch by Hubert Tong. 
[SVN r77638]
 2012-03-30 01:21:13 +00:00
Steven Watanabe
8db046a1f4 Make list_new take one argument. Rename the two argument form to list_push_back. 
[SVN r77522]
 2012-03-24 23:03:10 +00:00
Steven Watanabe
161a3c9462 Remove argument_list. It's unused. 
[SVN r77521]
 2012-03-24 22:15:42 +00:00
Steven Watanabe
daf90f7f79 Make sure that the arguments of built-in rules are actually checked. 
[SVN r77519]
 2012-03-24 21:49:12 +00:00
Steven Watanabe
888393c652 Merge Python function support into function.c 
[SVN r77518]
 2012-03-24 21:07:49 +00:00
Steven Watanabe
0b3b574817 Start cleaning up the arg_list hacks. Make the argument list part of the function instead of part of the rule. Don't var expand formal arguments. 
[SVN r77515]
 2012-03-24 18:53:21 +00:00
Steven Watanabe
f64a17d8b9 Change make to take a LIST *. There's no good reason to create a separate array. 
[SVN r77511]
 2012-03-23 22:55:50 +00:00
Steven Watanabe
3599b1926b Remove the first argument of list_copy, since it's always L0. 
[SVN r77510]
 2012-03-23 22:30:33 +00:00
Steven Watanabe
4f505e4a14 Fix memory leak 
[SVN r77503]
 2012-03-23 17:12:46 +00:00
Steven Watanabe
94925e3069 Fix memory leak. 
[SVN r77502]
 2012-03-23 16:52:48 +00:00
Steven Watanabe
74e780ea50 Fix annoying gcc warning. 
[SVN r77501]
 2012-03-23 16:40:04 +00:00
Steven Watanabe
5eec7d95fe Optimize variable lookup to bypass hash for constant variable names. 
[SVN r77491]
 2012-03-22 23:26:48 +00:00
Steven Watanabe
60d66ca888 Use dynamic arrays instead of linked lists. Reduces memory use about 10% overall. 
[SVN r77408]
 2012-03-19 18:17:36 +00:00
K. Noel Belcourt
a65106928d Eliminate uninitialized memory problem with sigemptyset. 
[SVN r77220]
 2012-03-04 17:23:44 +00:00
K. Noel Belcourt
584a0636af Use c, not c++, comments. 
[SVN r77198]
 2012-03-04 05:59:29 +00:00
K. Noel Belcourt
f38f7f06d1 Undo previous bad commit. 
[SVN r77197]
 2012-03-04 05:58:05 +00:00
K. Noel Belcourt
9e86e2fbb5 Use sigemptyset to initialize sa_mask instead of zero. 
[SVN r77193]
 2012-03-03 22:49:23 +00:00
K. Noel Belcourt
baa8a64459 Initialize variable in sigset_t. Thanks Steve 
for pointing this out.



[SVN r77192]
 2012-03-03 22:31:51 +00:00
Steven Watanabe
0ec332fea3 Fix alias violations in hash. 
[SVN r77190]
 2012-03-03 21:46:53 +00:00
Steven Watanabe
ed5f6b8bcf Downgrade strict syntax checks to a warning for backwards compatibility. 
[SVN r77093]
 2012-02-22 21:06:16 +00:00
K. Noel Belcourt
e199caba76 Protect unix specific structures. 
[SVN r77012]
 2012-02-14 02:17:51 +00:00
K. Noel Belcourt
3c7724acc5 Revert 77005, biffed the revert of 76862. 
[SVN r77009]
 2012-02-13 20:08:20 +00:00
K. Noel Belcourt
11f03ee4cd Revert commit 76862. This causes problems for a 
tester that we need to track down.



[SVN r77005]
 2012-02-13 16:36:00 +00:00