2
0
mirror of https://github.com/boostorg/build.git synced 2026-02-14 00:32:11 +00:00
Commit Graph

622 Commits

Author SHA1 Message Date
Vladimir Prus
7df1f2f4af Kill 'engine/src', moving all content to 'engine'.
[SVN r68592]
2011-02-01 08:16:47 +00:00
Vladimir Prus
e1f1c99d2f Move engine tests. Remove some releng bits that were useful only for
standardlone engine.


[SVN r68591]
2011-02-01 08:08:56 +00:00
Vladimir Prus
413e511014 Remove mistakenly added usage of Unix headers everywhere
[SVN r68475]
2011-01-27 16:25:26 +00:00
Vladimir Prus
307030a259 Commit the "don't stop when configure test fail" fix.
It appears to be only comitted to the release branch.


[SVN r68442]
2011-01-25 19:51:05 +00:00
Beman Dawes
dc832c90b6 tell boost inspect to ignore this directory
[SVN r67681]
2011-01-05 14:07:10 +00:00
Vladimir Prus
1282611bc0 Validate the parameter to -j.
Thanks to Michael Caisse for the bug report.


[SVN r67277]
2010-12-16 17:26:01 +00:00
K. Noel Belcourt
721f88717c Revert previous commit that changed how -l timeout is handled.
Decided that wasn't the best approach and am reverting the code
to the original source.



[SVN r66921]
2010-11-30 21:25:06 +00:00
K. Noel Belcourt
544a50bb11 Set bjam timeout to maximum of -l options on command line.
Patch to ensure that maximum value is selected when there's
multiple -l timeout options.



[SVN r66705]
2010-11-23 22:01:27 +00:00
K. Noel Belcourt
a9c45135e5 Fix poor parallel throughput on larger SMP systems.
When select() returns with data on one or more file descriptors, 
we use fread() to read data from the live descriptors.  The problem 
is that these are blocking file descriptors so most of the time 
bjam is waiting, it's actually waiting in fread(), rather than 
waiting in select().

There are two possible patches: one is to just call fread() a single  
time (not inside a loop) or we can make the file descriptors non- 
blocking.  It's more efficient to make the descriptors non-blocking  
as this allows us to read all data on a descriptor each time select()  
returns.  The first approach would not permit us to read all the data  
on a descriptor (only as much as fits into our buffer).

I tested this patch on Suse, Redhat, and Darwin.



[SVN r66650]
2010-11-20 00:15:41 +00:00
Bryce Adelstein-Lelbach
97eeccd165 Added comment suggested by volodya.
[SVN r66456]
2010-11-08 17:05:55 +00:00
Bryce Adelstein-Lelbach
754bc58425 Spoke to volodya about this; these changes add support for bootstrapping the
Boost.Build system with clang. I also fixed a few things in the clang-linux
toolchain;

  - Clang doesn't support -pg.
  - Adding -O0 to turn optimization off is silly, because it is set to zero by
    default if omitted on the command line. The compiler will whine about
    unused options if you later add another -O flag. The Boost.Build UTF
    interface adds a second -O0, so you end up with two redundant command line
    options and a warning from bjam/clang for every invocation of the compiler
    when using the test framework.



[SVN r66441]
2010-11-07 19:51:05 +00:00
Vladimir Prus
3e385dd38d Handle \t escape sequence.
Patch from Alexey Nikitin.


[SVN r66412]
2010-11-06 15:31:12 +00:00
Daniel James
c44506a7e3 Link fixes.
[SVN r66273]
2010-10-30 14:32:50 +00:00
Jürgen Hunold
331e379999 Fix: windows does not have wait(), use own macros instead.
[SVN r65319]
2010-09-06 14:14:09 +00:00
Vladimir Prus
4e316485e2 Add 'strip-eol' option to SHELL. Also, return actual exit status.
Previously, with the 'exit-code' option SHELL would return raw
value from waitXXX, with exit status shifted 8 bits to the left.

Fixes #4470.


[SVN r65232]
2010-09-04 10:10:34 +00:00
Steven Watanabe
2cd8cd3855 Supress -q for configuration checks.
[SVN r64861]
2010-08-17 16:49:44 +00:00
K. Noel Belcourt
45eafa3f5d Add bjam build for intel-darwin (Intel on Darwin).
[SVN r64647]
2010-08-06 18:44:15 +00:00
Vladimir Prus
1181c16034 Handle list of instances with __jam_repr__ method returned from Python.
This re-fixes test/inline.py broken by 'lib' changes.


[SVN r64634]
2010-08-06 11:22:42 +00:00
Vladimir Prus
c411fb609b Port build/configure.jam.
Also allow to expose Python class to Jam, which fixes
tag.py and inline.py testcases.


[SVN r64610]
2010-08-05 06:22:58 +00:00
Vladimir Prus
6267b550a0 Redo naming of actions again. Return values from Python to Jamfiles.
[SVN r64432]
2010-07-29 11:37:07 +00:00
Vladimir Prus
77d60028bb Various fixes
[SVN r64426]
2010-07-29 05:35:32 +00:00
Vladimir Prus
45f13ef7d8 Fix calling Python function taking zero parameters.
[SVN r64297]
2010-07-23 12:20:22 +00:00
Vladimir Prus
9ce9f999ab Compile on cygwin.
[SVN r64207]
2010-07-20 19:18:06 +00:00
K. Noel Belcourt
58ca782e5e Replace C++ (//) comments with C block comments (/* ... */).
This is one problem killing Itanium builds. One more to go.



[SVN r64204]
2010-07-20 17:13:19 +00:00
K. Noel Belcourt
c6d3de73a2 Fix typo stdrup with strdup.
This typo was in code section conditional on Sun
and the Sun nightly builds broke making this easy
to track down.



[SVN r64203]
2010-07-20 16:26:40 +00:00
Vladimir Prus
7255b1f5db Remove now-empty Boost.Jam docs directory.
[SVN r64190]
2010-07-20 08:23:53 +00:00
Vladimir Prus
398330e35e Move Boost.Jam docs to where Boost.Build docs are.
[SVN r64189]
2010-07-20 08:22:18 +00:00
Vladimir Prus
1cb68d9190 Perform bjam-relative search for Boost.Build proper.
This allows to make installed Boost.Build work without
setting any environment variables or config files.


[SVN r64165]
2010-07-19 19:32:32 +00:00
Vladimir Prus
e83d89e761 New builtin SELF_PATH
[SVN r64159]
2010-07-19 14:50:16 +00:00
Vladimir Prus
72a0c82fcd Implemented a mechanism for parameter translation from bjam to python
signatures.

This allows to expose the same interface to Jamfiles as before, e.g.

     rule toolset.flags ( rule variable condition * : values * )

and use natural interface in the Python side, where, 'rule', 'variable',
'condition' and 'values' are parameters, and the first two of them
will get string, not list of strings.


[SVN r64158]
2010-07-19 13:20:51 +00:00
Vladimir Prus
85352e0f67 Fix memory leak.
Apparently, PyList_Append does not steal reference, after all.


[SVN r64151]
2010-07-19 08:58:05 +00:00
Vladimir Prus
0f77dbdae2 Move tools/jam into tools/build/v2/engine
[SVN r64010]
2010-07-14 10:37:12 +00:00
Vladimir Prus
f137d18839 When no target is requested to be built via UPDATE, return status of last UPDATE_NOW.
[SVN r62635]
2010-06-09 07:12:59 +00:00
K. Noel Belcourt
61f6e8ee1f Remove non-blocking on child's pipe file descriptors.
Dan Gohman asked that we not force child's pipe file
descriptors to non-blocking so this commit removes
those fcntls calls.  I tested this change on Darwin,
Linux and Solaris and saw no adverse impact to nightly
build behavior.



[SVN r61961]
2010-05-14 04:06:03 +00:00
Rene Rivera
4eb744590e Bump bjam to 3.1.19.
[SVN r60778]
2010-03-22 21:04:59 +00:00
Rene Rivera
b372169a1a Add LIBS to the env vars list as it's also use in the cc toolset.
[SVN r60773]
2010-03-22 19:33:19 +00:00
Rene Rivera
89a530f5d6 Fix possible problems when trailing or leading whitespace is added to the env vars used, for example CC and CFLAGS when using the cc toolset. Even though I can't get compilation to fail even with the extra whitespace during building it does seem to fail for some compilers that don't ignore the whitespace arguments. (Fixes #3630)
[SVN r60768]
2010-03-22 15:26:45 +00:00
Vladimir Prus
8fe6def153 Use Python 2.6, if available
[SVN r60764]
2010-03-22 09:57:29 +00:00
Vladimir Prus
94f588650a Implement PRECIOUS builtin.
Note that existing code already tried to assume that the 'together'
flag means precious, but as described at:
http://public.perforce.com:8080/@md=d&cd=//public/jam/src/&cdf=//public/jam/src/make1.c&c=klD@/1494?ac=10
this was not really wrong.

Now, PRECIOUS makes the target never removed. I plan to extend it so that
user can control what exit codes cause the target to be removed.


[SVN r60157]
2010-03-04 23:02:57 +00:00
Vladimir Prus
2265402fc0 Fix UPDATE_NOW declaration
[SVN r59866]
2010-02-24 12:18:49 +00:00
Vladimir Prus
f2b25c5753 Add 'ignore -n' option to BUILTIN_UPDATE_NOW.
[SVN r59865]
2010-02-24 12:08:58 +00:00
Vladimir Prus
b170b53c38 Various new builtins.
[SVN r59829]
2010-02-21 23:33:13 +00:00
K. Noel Belcourt
20094eaacc This patch ensures we close both file descriptors following the dup2 calls.
This is mostly for completeness but also because I've occasionally noticed 
output isn't flushed from a terminated process (when it is killed for exceeding 
the time limit).  I'm sure this won't break anything and am slightly hopeful 
that it might ensure output is flushed before the terminating process is killed.



[SVN r58785]
2010-01-07 05:57:33 +00:00
Vladimir Prus
3910b10a57 Fix warnings. Closes #3656.
Patch from Mateusz Loskot.


[SVN r58011]
2009-11-28 17:46:46 +00:00
Vladimir Prus
73406e3691 Allow Boost.Build to control keep-going behaviour
[SVN r57978]
2009-11-27 19:57:44 +00:00
Vladimir Prus
5775cb918a Support more parameters to rules.
Fixes #2552.
Patch from Jonathan Biggar.


[SVN r57536]
2009-11-10 07:21:03 +00:00
Vladimir Prus
3e2fbb7677 Suppress warning.
[SVN r57403]
2009-11-05 08:05:20 +00:00
Vladimir Prus
d1b638f653 Don't mutate input to SPLIT_BY_CHARACTERS.
[SVN r56204]
2009-09-15 06:03:23 +00:00
Vladimir Prus
07571bbe5c Implement SPLIT_BY_CHARACTERS.
This is much faster than regex.split if splitting by a single char,
or by a set of one-char delimiters.


[SVN r56191]
2009-09-14 16:57:57 +00:00
Vladimir Prus
5e75949317 If command fails, print its output even for -d0.
[SVN r55042]
2009-07-20 10:10:19 +00:00