Martin Quinson [Sun, 24 Jan 2016 08:15:56 +0000 (09:15 +0100)]
tidyify a large XML file
Martin Quinson [Sun, 24 Jan 2016 08:15:24 +0000 (09:15 +0100)]
remove big useless deployment files
Martin Quinson [Sat, 23 Jan 2016 21:19:55 +0000 (22:19 +0100)]
Compile trace_mgr with g++. Next step: objectification
Martin Quinson [Sat, 23 Jan 2016 21:35:41 +0000 (22:35 +0100)]
explain in comment why we don't even test pthread on windows
Martin Quinson [Sat, 23 Jan 2016 21:17:45 +0000 (22:17 +0100)]
kill some more dead code from the C API to surf
Martin Quinson [Sat, 23 Jan 2016 21:06:52 +0000 (22:06 +0100)]
[java] do not load winpthread-1 on windows
- We made sure that we don't depend on that lib with the previous
commit.
- This makes things really simpler in the cmake, as we need only our
own libraries, that are the same on all platforms. No more cmake
conditionals that I fail to write for windows, no more sequencing
that fail with parallel builds.
- While I was at it, I did some reordering in the Java.cmake to have
related parts together.
Martin Quinson [Sat, 23 Jan 2016 20:53:29 +0000 (21:53 +0100)]
do not even test pthread on windows
This way, the Java library won't have any dependency on the
winpthread-1.dll library provided by mingw32-64.
Less dependencies, less problems.
Martin Quinson [Sat, 23 Jan 2016 09:23:21 +0000 (10:23 +0100)]
add some advertisement for Simulacrum
Frederic Suter [Fri, 22 Jan 2016 21:40:02 +0000 (22:40 +0100)]
this file is not here anymore
Frederic Suter [Fri, 22 Jan 2016 21:37:12 +0000 (22:37 +0100)]
deduplication and tidying
Martin Quinson [Fri, 22 Jan 2016 21:03:41 +0000 (22:03 +0100)]
cosmetics
Martin Quinson [Fri, 22 Jan 2016 20:58:45 +0000 (21:58 +0100)]
we will never distribute a windows installer again
We would need to build with Visual C for that (because C compiler are
not compatible on Windows), and this proved to be too hard for me.
Frederic Suter [Fri, 22 Jan 2016 20:28:38 +0000 (21:28 +0100)]
remove these huge generated files used with gtnets
Martin Quinson [Fri, 22 Jan 2016 20:39:20 +0000 (21:39 +0100)]
dedupplicate the platform files in the archive
Martin Quinson [Fri, 22 Jan 2016 20:39:00 +0000 (21:39 +0100)]
Do not look for the files that I just removed
Martin Quinson [Fri, 22 Jan 2016 20:13:56 +0000 (21:13 +0100)]
kill large unused useless dumb deployment files
Frederic Suter [Fri, 22 Jan 2016 19:34:08 +0000 (20:34 +0100)]
coverage madness in examples
Martin Quinson [Fri, 22 Jan 2016 17:16:53 +0000 (18:16 +0100)]
[jenkins] disable coverage of MC (libdw too ancient on slave)
Martin Quinson [Fri, 22 Jan 2016 17:07:40 +0000 (18:07 +0100)]
[java] dont even try to load boost_context
Martin Quinson [Fri, 22 Jan 2016 15:59:32 +0000 (16:59 +0100)]
kill dead code in the C interface to surf
Every surf user should use the C++ interface anyway
(+ useless cosmetics not worthing a commit on their own)
Frederic Suter [Fri, 22 Jan 2016 15:49:33 +0000 (16:49 +0100)]
plug a leak when ptask_LO7 is used
Frederic Suter [Fri, 22 Jan 2016 14:31:05 +0000 (15:31 +0100)]
improve coverage in teshsuite
Martin Quinson [Fri, 22 Jan 2016 15:28:28 +0000 (16:28 +0100)]
do not include libboost-context in the jar
- This is difficult because we may need to include libboost-chrono or
other libraries.
- This breaks the tests on some systems because of messages similar to
| OpenJDK Client VM warning: You have loaded library
| /tmp/simgrid-java7021501405032421762/libboost_context.so which might
| have disabled stack guard. The VM will try to fix the stack guard now.
- This make it difficult to build the jar on windows while allowing
parallel build.
So don't include the external libraries in the jar at all. We will
have to generate the native libraries with no dependencies for the
stable releases.
Martin Quinson [Fri, 22 Jan 2016 15:15:50 +0000 (16:15 +0100)]
typo in comments
Martin Quinson [Fri, 22 Jan 2016 15:13:12 +0000 (16:13 +0100)]
kill dead code. Simgrid internals are Pure Magic
degomme [Fri, 22 Jan 2016 14:40:15 +0000 (15:40 +0100)]
add missing call in host (available in MSG, used in lua)
Martin Quinson [Fri, 22 Jan 2016 13:50:33 +0000 (14:50 +0100)]
plug a memleak
Frederic Suter [Fri, 22 Jan 2016 13:00:26 +0000 (14:00 +0100)]
completely useless commit (reindent repetitive xml)
degomme [Fri, 22 Jan 2016 12:50:51 +0000 (13:50 +0100)]
Revert "[jenkins] also exclude generated files from the coverage"
This reverts commit
b218ac72df339ff0d8dc9e063c4820a4fee831eb.
degomme [Fri, 22 Jan 2016 12:50:18 +0000 (13:50 +0100)]
Enable everything I can think of for coverage
Frederic Suter [Fri, 22 Jan 2016 12:11:00 +0000 (13:11 +0100)]
change argument testing method (100% coverage)
degomme [Fri, 22 Jan 2016 11:53:11 +0000 (12:53 +0100)]
requalify tesh
degomme [Fri, 22 Jan 2016 11:47:30 +0000 (12:47 +0100)]
fix compilation with jedule and limited by latency
Martin Quinson [Fri, 22 Jan 2016 10:51:51 +0000 (11:51 +0100)]
remove two broken tests. Who use ptask anyway? ;)
Martin Quinson [Fri, 22 Jan 2016 10:51:18 +0000 (11:51 +0100)]
revalidate tesh files when maestro is given a process name in the logs
Martin Quinson [Fri, 22 Jan 2016 09:53:37 +0000 (10:53 +0100)]
Kill another GRAS leftover: SG/RL thread portability layer
Also, maestro's reported name is now "maestro" instead of "". This
breaks a lot of tesh files that I'll fix in the next commit for clarity.
Christian Heinrich [Thu, 21 Jan 2016 11:06:59 +0000 (12:06 +0100)]
[Lua] Removed MSG from Lua
Christian Heinrich [Thu, 21 Jan 2016 10:59:26 +0000 (11:59 +0100)]
[LUA] Removed another outdated lua function
degomme [Fri, 22 Jan 2016 10:22:09 +0000 (11:22 +0100)]
try to fix build with jedule/latency bound
degomme [Thu, 21 Jan 2016 10:40:27 +0000 (11:40 +0100)]
changelog update
Martin Quinson [Thu, 21 Jan 2016 20:14:04 +0000 (21:14 +0100)]
kill another GRAS leftover: xbt_peer_t
Martin Quinson [Thu, 21 Jan 2016 19:53:48 +0000 (20:53 +0100)]
some more renamings in the traces
Martin Quinson [Thu, 21 Jan 2016 19:29:15 +0000 (20:29 +0100)]
[jenkins] also exclude generated files from the coverage
Martin Quinson [Thu, 21 Jan 2016 18:18:39 +0000 (19:18 +0100)]
Merge branch 'master' of github.com:mquinson/simgrid
Martin Quinson [Thu, 21 Jan 2016 18:09:48 +0000 (19:09 +0100)]
Did I broke travis?
Martin Quinson [Thu, 21 Jan 2016 15:07:28 +0000 (16:07 +0100)]
First round of refactoring in the event trace stuff
- Rename various data types and function parameters to make their
purpose clear
- More to come: A class hierarchy is to be discovered in these messy bits
Martin Quinson [Thu, 21 Jan 2016 12:25:37 +0000 (13:25 +0100)]
Merge pull request #52 from EtienneM/patch-1
Typo in Task javadoc
EtienneM [Thu, 21 Jan 2016 11:06:58 +0000 (12:06 +0100)]
Documentation for parallel tasks
EtienneM [Thu, 21 Jan 2016 10:21:14 +0000 (11:21 +0100)]
Typo in Task javadoc
degomme [Thu, 21 Jan 2016 10:06:13 +0000 (11:06 +0100)]
exclude teshsuite folder from coverage reports, as it's not meaningful
Martin Quinson [Thu, 21 Jan 2016 10:00:48 +0000 (11:00 +0100)]
[xbt] Kill the queue datacontainer: it made more sense with GRAS
degomme [Thu, 21 Jan 2016 09:49:36 +0000 (10:49 +0100)]
requalify soome tesh files
Martin Quinson [Thu, 21 Jan 2016 09:45:20 +0000 (10:45 +0100)]
cosmetics in teshsuite/msg/host_on_off -- don't ask why
Martin Quinson [Thu, 21 Jan 2016 02:28:20 +0000 (03:28 +0100)]
woops revalidate some tesh files
Martin Quinson [Thu, 21 Jan 2016 01:32:04 +0000 (02:32 +0100)]
ignore more generated files
degomme [Thu, 21 Jan 2016 09:07:56 +0000 (10:07 +0100)]
Using openMPI selector in this case possibly led to issues, so use single implems instead
Martin Quinson [Thu, 21 Jan 2016 01:20:40 +0000 (02:20 +0100)]
Also run the energy tests with ptask_L07 model.
ptask_L07 needs to shut up when asked (when xbt_cfg is set to critical)
Martin Quinson [Thu, 21 Jan 2016 01:20:08 +0000 (02:20 +0100)]
allow to add log settings to the executed commands
Martin Quinson [Thu, 21 Jan 2016 00:45:32 +0000 (01:45 +0100)]
Document some important changes introduced in this release
Martin Quinson [Thu, 21 Jan 2016 00:31:40 +0000 (01:31 +0100)]
fix make distcheck
Martin Quinson [Thu, 21 Jan 2016 00:21:13 +0000 (01:21 +0100)]
integrate Adrien's example for Java VM energy as a tesh test
degomme [Thu, 21 Jan 2016 00:13:46 +0000 (01:13 +0100)]
enable mpich3 tests for coverage reports on ci
Christian Heinrich [Thu, 21 Jan 2016 00:13:17 +0000 (01:13 +0100)]
[LUA] Fixed compile time warnings, 2/2
Christian Heinrich [Thu, 21 Jan 2016 00:11:12 +0000 (01:11 +0100)]
[LUA] Fixed compile time warnings
degomme [Wed, 20 Jan 2016 22:59:09 +0000 (23:59 +0100)]
add a test for IB network model as well
Christian Heinrich [Wed, 20 Jan 2016 22:39:12 +0000 (23:39 +0100)]
[CMAKE] Fixed build system by removing lua crap
Christian Heinrich [Wed, 20 Jan 2016 22:26:40 +0000 (23:26 +0100)]
[DOC] Fixed even more errors.
Christian Heinrich [Wed, 20 Jan 2016 22:07:23 +0000 (23:07 +0100)]
[Gitignore] Blacklisted mpich3-test files.
Christian Heinrich [Wed, 20 Jan 2016 22:01:51 +0000 (23:01 +0100)]
[Docs] Fixed doc generation errors.
degomme [Wed, 20 Jan 2016 21:59:54 +0000 (22:59 +0100)]
drop n-ary bcast/barrier that was not actually working as a barrier.
Default is now ompi selector barrier/bcast.
degomme [Wed, 20 Jan 2016 21:44:25 +0000 (22:44 +0100)]
Some weird codes actually call MPI_Init from C, then call MPI functions from fortran
In this case we did not have fortran specific structures initialized, thus segfault.
This should help
Frederic Suter [Wed, 20 Jan 2016 17:58:27 +0000 (18:58 +0100)]
more useless code trimmed
degomme [Wed, 20 Jan 2016 18:32:53 +0000 (19:32 +0100)]
another fix to avoid issues with .f or .F in f77
Christian Heinrich [Wed, 20 Jan 2016 17:16:57 +0000 (18:16 +0100)]
[Lua] Removed remains from older codebase
Christian Heinrich [Wed, 20 Jan 2016 17:14:30 +0000 (18:14 +0100)]
[Lua] Removed orphaned comments
Christian Heinrich [Wed, 20 Jan 2016 17:03:35 +0000 (18:03 +0100)]
[Lua] Added new 'dump()' debug function for lua
This function will help users to easily dump their tables
by calling 'simgrid.dump(table)' in Lua. This is not really
an elaborate solution (no recursion handling), but it's OK
for the moment
Christian Heinrich [Wed, 20 Jan 2016 17:02:27 +0000 (18:02 +0100)]
[Lua] Removed lua_utils.c for lua_debug.c
Christian Heinrich [Tue, 19 Jan 2016 23:35:56 +0000 (00:35 +0100)]
[Lua] Further removal of unnecessary lua glue code
Christian Heinrich [Tue, 19 Jan 2016 23:10:45 +0000 (00:10 +0100)]
[Tests][Lua] Fixed wrong Lua platforms test
Christian Heinrich [Tue, 19 Jan 2016 22:59:38 +0000 (23:59 +0100)]
[Lua] Removed lua simulation support
Christian Heinrich [Tue, 19 Jan 2016 21:22:25 +0000 (22:22 +0100)]
[Lua] Renamed host 'power' to 'speed' for hosts
Christian Heinrich [Tue, 19 Jan 2016 16:50:00 +0000 (17:50 +0100)]
[Contrib] Deleted contrib/network_model
Christian Heinrich [Tue, 19 Jan 2016 11:22:53 +0000 (12:22 +0100)]
[DOC] Added notes about SMPI_SAMPLE macros
Christian Heinrich [Mon, 18 Jan 2016 21:32:36 +0000 (22:32 +0100)]
[Doc] Fixed wrong mandatory flag
degomme [Wed, 20 Jan 2016 17:13:06 +0000 (18:13 +0100)]
Add test for smpi tracing with only -trace, and no other option
degomme [Wed, 20 Jan 2016 17:09:48 +0000 (18:09 +0100)]
fix potential segfault happening when only -trace is provided to smpi
Ad(rien) L [Wed, 20 Jan 2016 17:03:00 +0000 (18:03 +0100)]
Merge branch 'master' of git://scm.gforge.inria.fr/simgrid/simgrid
Ad(rien) L [Wed, 20 Jan 2016 17:02:49 +0000 (18:02 +0100)]
add a naive vm energy example - Ad(rien)
Frederic Suter [Wed, 20 Jan 2016 15:05:35 +0000 (16:05 +0100)]
Starting the SimDag revolution: Easter trimming
+ Loose the SEQUENTIAL_ACCESS_MODE in the process
Gabriel Corona [Wed, 20 Jan 2016 15:11:01 +0000 (16:11 +0100)]
Explicit instanciation of Extendable<simgrid::s4u::Host>
It seems the `deleters_` static variable (of a template class) is not
uniquified on some systems (where the symbols are `STB_GLOBAL`) when
using Java loadLibrary leading to the JAVA_HOST_LEVEL being the same
as another level.
On some systems, the symbols are `STB_GNU_UNIQUE` which purpose is to
avoid this kind of behaviour and the bug does not happen.
By using explicit instanciation, the `deleters_` variable is not
allocated in each shared-object but only in `libsimgrid.so`.
degomme [Wed, 20 Jan 2016 15:49:04 +0000 (16:49 +0100)]
Add a test for smpi+different kinds of clusters (torus/fat_tree/flat/two clusters)
degomme [Wed, 20 Jan 2016 11:26:55 +0000 (12:26 +0100)]
Hmm, that was a lie.
I can't remember why we needed that.
Martin Quinson [Wed, 20 Jan 2016 09:47:45 +0000 (10:47 +0100)]
more argument checking
Martin Quinson [Tue, 19 Jan 2016 17:16:28 +0000 (18:16 +0100)]
Try to activate coverity for simgrid
Frederic Suter [Wed, 20 Jan 2016 08:49:41 +0000 (09:49 +0100)]
mv private function header to the right place
Frederic Suter [Tue, 19 Jan 2016 16:33:05 +0000 (17:33 +0100)]
SD_link_get_list -> sd_link_list
Frederic Suter [Tue, 19 Jan 2016 15:05:06 +0000 (16:05 +0100)]
yes, why?
degomme [Wed, 20 Jan 2016 00:27:06 +0000 (01:27 +0100)]
be careful never to change the case of the extension in fortran ...
Yes, it matters because one forces the C preprocessor to pass, and not the other.
degomme [Tue, 19 Jan 2016 23:34:21 +0000 (00:34 +0100)]
Don't tell the user to use an option his simgrid soes not support (but tease him).
Signed-off-by: degomme <augustin.degomme@unibas.ch>
degomme [Tue, 19 Jan 2016 22:49:58 +0000 (23:49 +0100)]
Make all predefined MPI_Datatype const
We encountered one MPI application that was using them as such and thus failing to build with SMPI.