f7346fc | Roberto Di Cosmo | 30 November 2017, 20:58:28 UTC | Merge pull request #65 from UnixJunkie/patch-1 same as in the opam repository | 30 November 2017, 20:58:28 UTC |
1855b87 | Roberto Di Cosmo | 30 November 2017, 20:57:20 UTC | Merge pull request #67 from UnixJunkie/rm_trailing_whitespaces remove trailing whitespaces | 30 November 2017, 20:57:20 UTC |
e8ad8a7 | Francois BERENGER | 30 November 2017, 00:59:54 UTC | remove trailing whitespaces | 30 November 2017, 00:59:54 UTC |
b030c05 | Roberto Di Cosmo | 20 September 2017, 11:39:59 UTC | First stab at adding explicit core mapping. | 28 November 2017, 08:36:40 UTC |
29c9d48 | Francois BERENGER | 27 November 2017, 06:30:19 UTC | same as in the opam repository added conf-aclocal dep to the opam file | 27 November 2017, 06:30:19 UTC |
699ed2a | Roberto Di Cosmo | 18 November 2017, 10:43:40 UTC | Merge pull request #64 from thierry-martinez/master Fix for OCaml 4.06 | 18 November 2017, 10:43:40 UTC |
20695e4 | Thierry Martinez | 18 November 2017, 10:20:41 UTC | Comment references to bytes | 18 November 2017, 10:20:41 UTC |
ddcab19 | Thierry Martinez | 18 November 2017, 00:19:57 UTC | Fix for OCaml 4.06 | 18 November 2017, 00:19:57 UTC |
c9b0ee7 | Roberto Di Cosmo | 24 February 2017, 00:29:58 UTC | Remove cruft after merge | 24 February 2017, 00:29:58 UTC |
314f1ad | Roberto Di Cosmo | 23 February 2017, 21:02:33 UTC | Merge pull request #54 from UnixJunkie/get_ncores added get_ncores, thanks to @UnixJunkie | 23 February 2017, 21:02:33 UTC |
ac94452 | Francois BERENGER | 23 February 2017, 16:44:37 UTC | Merge branch 'master' into get_ncores | 23 February 2017, 16:44:37 UTC |
743ec49 | Roberto Di Cosmo | 23 February 2017, 00:42:16 UTC | Merge pull request #55 from UnixJunkie/code_factorization Gc.compact needs to be called only inside of spawn_many | 23 February 2017, 00:42:16 UTC |
400fd71 | Roberto Di Cosmo | 23 February 2017, 00:36:09 UTC | Merge pull request #56 from UnixJunkie/get_rank added get_rank so that worker processes can know at which rank | 23 February 2017, 00:36:09 UTC |
8204dbb | Francois BERENGER | 22 February 2017, 18:54:15 UTC | added get_rank so that worker processes can know at which rank they were forked out of the master process | 22 February 2017, 18:54:15 UTC |
f3643a0 | Francois BERENGER | 22 February 2017, 18:19:24 UTC | Gc.compact needs to be called only inside of spawn_many | 22 February 2017, 18:19:24 UTC |
261f9da | Francois BERENGER | 22 February 2017, 18:02:56 UTC | added get_ncores | 22 February 2017, 18:02:56 UTC |
af9691e | Roberto Di Cosmo | 02 November 2016, 14:18:39 UTC | Merge pull request #50 from diml/master handle EINTR errors in wait_for_pids | 02 November 2016, 14:18:39 UTC |
52c1539 | Jeremie Dimino | 02 November 2016, 14:05:44 UTC | handle EINTR errors in wait_for_pids | 02 November 2016, 14:05:44 UTC |
fa5c3dd | Roberto Di Cosmo | 02 November 2016, 13:00:06 UTC | Merge pull request #49 from diml/master Prevent at_exit handlers registered in the parent from running in the child | 02 November 2016, 13:00:06 UTC |
c8aab66 | Jérémie Dimino | 02 November 2016, 11:44:14 UTC | Prevent at_exit handlers registered in the parent from running in the child | 02 November 2016, 11:44:14 UTC |
b1abe53 | Roberto Di Cosmo | 26 October 2016, 15:51:57 UTC | Merge pull request #48 from diml/master Only wait for the children started by parmap | 26 October 2016, 15:51:57 UTC |
99d9d23 | Jeremie Dimino | 26 October 2016, 10:14:13 UTC | Only wait for the children started by parmap Instead of calling Unix.wait n times, call Unix.waitpid on all the pids of the children parmap started. | 26 October 2016, 10:14:13 UTC |
10e2437 | Roberto Di Cosmo | 05 June 2016, 19:31:59 UTC | Merge pull request #45 from jeffmahoney/master can_redirect: don't treat mkdir returning EEXIST as an error | 05 June 2016, 19:31:59 UTC |
a1e92fe | Jeff Mahoney | 05 June 2016, 00:21:04 UTC | can_redirect: don't treat mkdir returning EEXIST as an error can_redirect's test for existance of the path followed by the mkdir is inherently racy. It's trivially possible to encounter a situation where the test for existance fails in two processes and one of them succeeds in creating the directory. Since both processes are looking to create and use the directory for the same purpose, we don't need to treat EEXIST as an error. | 05 June 2016, 00:21:04 UTC |
7598fb9 | Gabriel Scherer | 08 May 2016, 13:36:58 UTC | add ./opam taken from opam-repository | 12 May 2016, 08:08:39 UTC |
5813de5 | Gabriel Scherer | 08 May 2016, 13:29:25 UTC | Makefile: also install .cmx files Since 4.03 (really for flambda), the native compiler is sensibly more sensitive to the presence of .cmx files. There is now a warning if the .cmx for a module is absent, unless its interface (.cmi) has been compiled with -opaque. | 12 May 2016, 08:08:39 UTC |
e550505 | Roberto Di Cosmo | 12 May 2016, 07:52:39 UTC | Add information on the build system. | 12 May 2016, 07:52:39 UTC |
8a39374 | Roberto Di Cosmo | 12 May 2016, 07:37:50 UTC | Proper handling of dependencies, thanks Gabriel Sherer for spotting this | 12 May 2016, 07:37:50 UTC |
970e49b | Roberto Di Cosmo | 11 May 2016, 15:07:32 UTC | Reverting one option that still breaks compilation in 4.02, leaving open to fix | 11 May 2016, 15:07:32 UTC |
219e3d0 | Roberto Di Cosmo | 11 May 2016, 14:34:13 UTC | Fix for 4.03.0 issue in linking setcore into parmap.cmxa We should find a more proper way of doing this. | 11 May 2016, 14:42:56 UTC |
2abc50e | Roberto Di Cosmo | 30 April 2016, 16:17:45 UTC | Clean up Changelog | 30 April 2016, 16:17:45 UTC |
ce28808 | Roberto Di Cosmo | 27 April 2016, 09:42:32 UTC | Merge pull request #41 from UnixJunkie/fixes_for_compilation_w4.03.0 Fixes for compilation w4.03.0 : thanks Francois. | 27 April 2016, 09:42:32 UTC |
d5d197c | Francois BERENGER | 27 April 2016, 09:23:49 UTC | corrected warning related to Error type | 27 April 2016, 09:23:49 UTC |
6551bab | Francois BERENGER | 27 April 2016, 09:21:33 UTC | mv setcore.mli setcore.ml | 27 April 2016, 09:21:33 UTC |
a9f3396 | Roberto Di Cosmo | 17 September 2015, 14:34:21 UTC | Bumped up version string | 17 September 2015, 14:34:21 UTC |
51c6666 | Jérémie Courrèges-Anglas | 17 September 2015, 13:24:21 UTC | Makefile.in portability fixes Good news: https://forge.ocamlcore.org/plugins/scmgit/cgi-bin/gitweb.cgi?p=ocaml-autoconf/ocaml-autoconf.git;a=commitdiff;h=9e0d03be521283b59e002d91422b7695825195e3 With a fresher ocaml.m4 & ./configure, the following diff allows one to automatically build the .cmxs on OpenBSD/amd64 (native dynlink supported) and to skip it on OpenBSD/sparc64 (non-native). | 17 September 2015, 13:37:17 UTC |
1cc6a04 | Roberto Di Cosmo | 17 September 2015, 13:35:32 UTC | Updated ocaml.m4 to version on master as of 17/9/2015 (not yet released on ocamlforge) | 17 September 2015, 13:35:32 UTC |
58f56f5 | Jérémie Courrèges-Anglas | 16 September 2015, 18:44:54 UTC | Makefile.in portability fixes Following configure.ac, two things in Makefile.in prevent me from using it verbatim. First, our cp(1) does not support the -a flag. I'd say that it could easily be replaced by -p. Second issue, parmap.a shouldn't be built on architectures that don't have native code support (ie. where ocamlopt is present and usable). There's another glitch but I'm afraid I can't fix it given the existing ocaml.m4: .cmxs plugins aren't supported on all native archs (for example OpenBSD/sparc). Here are two patches for the two issues. Cheers, >From 54bd6c21e9f56f58be29c8ef923e3a0ca2c32f15 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?J=C3=A9r=C3=A9mie=20Courr=C3=A8ges-Anglas?= <jca@wxcvbn.org> Date: Wed, 16 Sep 2015 20:26:05 +0200 Subject: [PATCH 1/2] Portability: use cp -p instead of -a | 17 September 2015, 10:39:17 UTC |
a276c4a | Jérémie Courrèges-Anglas | 16 September 2015, 11:41:00 UTC | Fix autoconf test with shells != bash Improve the portability of the test for ocamlbuild -use-ocamlfind. * shell redirections with "&> /dev/null" and the test(1) operator "==" are not portable. Fix both issues and simplify the test while here. With this, ./configure on OpenBSD correctly uses Makefile.in and not Makefile_3.11.in. | 16 September 2015, 12:42:46 UTC |
642dc2b | Roberto Di Cosmo | 23 April 2015, 18:26:51 UTC | Improved mkdwn | 23 April 2015, 18:26:51 UTC |
5b50969 | Roberto Di Cosmo | 23 April 2015, 18:26:19 UTC | Improved mkdwn | 23 April 2015, 18:26:19 UTC |
6fb0eb0 | Roberto Di Cosmo | 23 April 2015, 18:24:42 UTC | Improved mkdwn | 23 April 2015, 18:24:42 UTC |
a9f297a | Roberto Di Cosmo | 23 April 2015, 18:23:39 UTC | Improved mkdwn | 23 April 2015, 18:23:39 UTC |
738eb7c | Roberto Di Cosmo | 23 April 2015, 18:22:00 UTC | Improved mkdwn | 23 April 2015, 18:22:00 UTC |
0da42ca | Roberto Di Cosmo | 23 April 2015, 18:19:10 UTC | Adding mkdwn vesion of REAMDE | 23 April 2015, 18:19:25 UTC |
715ae7a | Roberto Di Cosmo | 09 March 2015, 13:58:48 UTC | Merge pull request #34 from nilsbecker/master os x specific Makefile | 09 March 2015, 13:58:48 UTC |
1c3c5d5 | nilsbecker | 03 March 2015, 15:54:58 UTC | Update Makefile.OSX | 03 March 2015, 15:54:58 UTC |
136f70d | nilsbecker | 03 March 2015, 15:48:43 UTC | Update and rename example to example/Makefile.OSX | 03 March 2015, 15:48:43 UTC |
942f90e | Doug Hogan | 09 February 2015, 07:57:40 UTC | Fix build error on systems without thread_policy | 09 February 2015, 10:42:28 UTC |
b0f230d | Roberto Di Cosmo | 08 January 2015, 09:58:47 UTC | Merge pull request #32 from xclerc/silence-warnings Silence warnings for unused variables. And remove finalize parameter from setup_children_chans for now. | 08 January 2015, 09:58:47 UTC |
752de8e | Xavier Clerc | 08 January 2015, 09:48:50 UTC | Suppress 'finalize' parameter from 'setup_children_chans' function. | 08 January 2015, 09:48:50 UTC |
f9540fe | Xavier Clerc | 08 January 2015, 09:48:17 UTC | Silence warnings for unused variables. | 08 January 2015, 09:48:17 UTC |
a7cd1f7 | Roberto Di Cosmo | 20 October 2014, 08:53:57 UTC | Bumped up version string | 20 October 2014, 08:53:57 UTC |
d6688ca | Roberto Di Cosmo | 12 October 2014, 16:39:07 UTC | added options to get warnings, as suggested by xclerc | 12 October 2014, 16:39:07 UTC |
599878e | Roberto Di Cosmo | 11 October 2014, 09:00:42 UTC | Merge pull request #28 from xclerc/master Init/finalize functions not called by parmapfold, parfold, and array_float_parmap. | 11 October 2014, 09:00:42 UTC |
997ca99 | Xavier Clerc | 10 October 2014, 09:10:20 UTC | Init/finalize functions not called by parmapfold, parfold, and array_float_parmap. | 10 October 2014, 09:10:20 UTC |
aa2270a | Roberto Di Cosmo | 16 May 2014, 16:40:36 UTC | Merge pull request #25 from UnixJunkie/more_ocamldoc some ocamldoc for init and finalize | 16 May 2014, 16:40:36 UTC |
ce335a7 | Francois Berenger | 16 May 2014, 02:39:24 UTC | some ocamldoc for init and finalize | 16 May 2014, 02:39:24 UTC |
9f1f03a | Roberto Di Cosmo | 15 May 2014, 14:59:44 UTC | fixed mandir | 15 May 2014, 14:59:44 UTC |
23166c3 | Roberto Di Cosmo | 15 May 2014, 14:32:40 UTC | Step up version in code | 15 May 2014, 14:32:40 UTC |
d803e2b | Roberto Di Cosmo | 15 May 2014, 14:31:47 UTC | Restore correct paths in Makefile | 15 May 2014, 14:31:47 UTC |
168a34e | Roberto Di Cosmo | 15 May 2014, 14:31:16 UTC | Step up version in code | 15 May 2014, 14:31:16 UTC |
85ac385 | Roberto Di Cosmo | 15 May 2014, 14:31:04 UTC | Step up version in code | 15 May 2014, 14:31:04 UTC |
cb5a4b8 | Roberto Di Cosmo | 12 May 2014, 11:23:33 UTC | Replaced multiple calls of finalize with Pervasives.at_exit; moved instances of init rigth after the fork | 12 May 2014, 11:23:33 UTC |
b523eb0 | Roberto Di Cosmo | 12 May 2014, 10:04:52 UTC | Updated utils.ml in tests to allow passing init functions | 12 May 2014, 10:09:13 UTC |
236137f | Roberto Di Cosmo | 12 May 2014, 09:56:23 UTC | Remove redirection flag, and introduce a redirect helper function | 12 May 2014, 10:09:07 UTC |
498782c | Roberto Di Cosmo | 12 May 2014, 09:28:30 UTC | Following advice from @UnixJunkie, initialisation function defaults to doing nothing. We need to expose a redirect function instead, helping who may want to use it without rewriting it from scratch. | 12 May 2014, 09:56:51 UTC |
dd9f089 | Roberto Di Cosmo | 10 May 2014, 13:56:47 UTC | Call init and finalize in the workers and in the finish function of setup_chan | 12 May 2014, 09:26:22 UTC |
c692d6a | Roberto Di Cosmo | 10 May 2014, 13:56:01 UTC | Added init and finalize parameters to the internal functions | 12 May 2014, 09:26:22 UTC |
13d9234 | Roberto Di Cosmo | 10 May 2014, 13:53:19 UTC | Adding init and finalize parameters to the main functions. | 12 May 2014, 09:26:22 UTC |
fa0ee6e | Roberto Di Cosmo | 10 May 2014, 12:16:53 UTC | Use redirect as default initialisation function | 12 May 2014, 09:26:10 UTC |
b3d9247 | Roberto Di Cosmo | 12 May 2014, 09:25:49 UTC | added init to mapper | 12 May 2014, 09:25:49 UTC |
1eaa992 | Roberto Di Cosmo | 10 May 2014, 08:44:26 UTC | The initialisation function will receive the number of the core it is running on. | 12 May 2014, 09:24:38 UTC |
b096fc7 | Roberto Di Cosmo | 10 May 2014, 08:17:25 UTC | Start adding initialisation and finalisation functions to each parallel combinator. | 12 May 2014, 09:24:27 UTC |
1fc192b | Roberto Di Cosmo | 12 May 2014, 09:23:42 UTC | Factor exception handling code | 12 May 2014, 09:23:42 UTC |
125cb4d | Roberto Di Cosmo | 26 February 2014, 22:12:20 UTC | Added code to get the number of available cores, and use it to compute default number of cores. | 26 February 2014, 22:12:20 UTC |
ee86243 | Roberto Di Cosmo | 02 September 2013, 14:47:10 UTC | Merge pull request #16 from pippijn/install-fixes fix make install to obey DESTDIR | 02 September 2013, 14:47:10 UTC |
54a1589 | Pippijn van Steenhoven | 25 July 2013, 10:37:50 UTC | don't check for camlp4, parmap doesn't use it fixes rdicosmo/parmap#17 | 25 July 2013, 11:25:48 UTC |
f245a88 | Pippijn van Steenhoven | 25 July 2013, 01:50:32 UTC | fix make install to obey DESTDIR | 25 July 2013, 01:50:32 UTC |
61f92a7 | Roberto Di Cosmo | 25 June 2013, 13:12:17 UTC | Added generation and installation of the manpage | 25 June 2013, 13:12:17 UTC |
1d23faa | Roberto Di Cosmo | 23 February 2013, 10:46:11 UTC | Merge branch 'master' of github.com:rdicosmo/parmap | 23 February 2013, 10:46:11 UTC |
f8821c3 | Roberto Di Cosmo | 23 February 2013, 10:45:20 UTC | Removed outdated instructions for OCaml 3.11 users (configure properly handles this case). | 23 February 2013, 10:45:20 UTC |
bf3d603 | Roberto Di Cosmo | 22 February 2013, 09:31:01 UTC | added interface for parmap_utils | 22 February 2013, 09:31:01 UTC |
c6a4a8c | Roberto Di Cosmo | 18 February 2013, 22:54:14 UTC | Removed stray myocamlbuild.* from installation | 18 February 2013, 22:54:14 UTC |
4dc6d3d | Roberto Di Cosmo | 18 February 2013, 22:43:04 UTC | Merge branch 'master' of github.com:rdicosmo/parmap | 18 February 2013, 22:43:04 UTC |
f081720 | Roberto Di Cosmo | 18 February 2013, 22:39:58 UTC | Merge pull request #12 from rixed/fix-link-bytecode Add required cclib to link with parmap.cma (thanks Cedric) | 18 February 2013, 22:39:58 UTC |
85aa0c5 | Cedric Cellier | 18 February 2013, 21:24:29 UTC | Add required cclib to link with parmap.cma When linking statically against parmap, one need to link with parmap_stubs library (since it's not recorded in the cma itself). Findlib can automatically add the required -cclib option. | 18 February 2013, 21:24:29 UTC |
00e81c8 | Roberto Di Cosmo | 18 February 2013, 16:36:04 UTC | Bumped up version to 1.0-rc2 | 18 February 2013, 16:36:04 UTC |
e873e1e | Samuel Hornus | 18 February 2013, 16:33:11 UTC | Fixed typo in code for Mac OS X core scheduling. | 18 February 2013, 16:33:11 UTC |
7596fdc | Roberto Di Cosmo | 18 February 2013, 15:36:43 UTC | Bumped up version to 1.0-rc1 | 18 February 2013, 15:36:43 UTC |
0ebc0a6 | Roberto Di Cosmo | 10 October 2012, 07:33:01 UTC | Add parmap_utils to the build system | 10 October 2012, 07:33:01 UTC |
5dfb222 | Roberto Di Cosmo | 09 October 2012, 13:44:41 UTC | Added parmap_utils to Makefile_3.11.in | 09 October 2012, 13:44:41 UTC |
6afce79 | Francois Berenger | 15 August 2012, 04:30:15 UTC | same indentation level in whole file | 15 August 2012, 04:30:15 UTC |
c8af916 | Francois Berenger | 15 August 2012, 04:28:44 UTC | debug became log_debug so that it is more explicit about what it does | 15 August 2012, 04:28:44 UTC |
52d6398 | Francois Berenger | 15 August 2012, 04:24:57 UTC | - some code was moved in parmap_utils.ml so now a Utils module is used explicitely - smarter use of backslash in long strings - rm some empty lines - better formatting in some parts | 15 August 2012, 04:24:57 UTC |
bd9bb29 | Francois Berenger | 15 August 2012, 04:08:41 UTC | added so that functions not directly related to parallelization in parmap.ml can move to this file | 15 August 2012, 04:08:41 UTC |
2c6a86d | Francois Berenger | 15 August 2012, 02:33:41 UTC | moved dead code into the comment it looks related to | 15 August 2012, 02:33:41 UTC |
9ca7112 | Francois Berenger | 15 August 2012, 02:07:44 UTC | info was renamed to log_error as previous name was misleading | 15 August 2012, 02:07:44 UTC |
567e9da | Francois Berenger | 15 August 2012, 02:01:34 UTC | emacs removed all trailing whitespaces | 15 August 2012, 02:01:34 UTC |
abb70e0 | Francois Berenger | 15 August 2012, 02:00:54 UTC | cut long lines (> 80 chars) | 15 August 2012, 02:00:54 UTC |