direct links to git repo for all applications
5 Mar 2014 19:53:54 UTC
Topic 197429
(moderation:
eg, for "Gamma-ray pulsar search #3" I'd like to know where I can start to dig into the code;
http://www.lsc-group.phys.uwm.edu/cgit/lalsuite/tree
is not intuitively to me.
direct links to git repo for all applications
)
That URL would be one to enter into a git command line, or git GUI, in order to clone/download the repository. Do you have experience with repo's, git and so forth ? What system are you using ?
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Ofc I have experiences with
)
Ofc I have experiences with repos - but I'd expect a working cgit (or similar) web interface too for a first quick view.
RE: http://www.lsc-group.ph
)
What precisely do you mean not to be "intuitively"?
LALSuite is a huge library, containing a large number of different codes that implement various algorithms and other parts of all Gravitational Wave searches that the LSC with currently more than 900 people contribute to. The search performed on Einstein@Home is just a small part of that effort and uses only a small fraction of that code. As a library it is rather meant to be complete (maintaining a minimum level of coherence) than to be intuitively "usable" (or browsable, or whatever). If you really want to dig into it, rather than browing the code I'd recommend to start with the doxygen documentation (generated nightly), beginning with the applications in LALApps.
The code for the Radio Pulsar search (BRP*) is indeed maintained in a git repository. But as there also is data in there e.g. about yet unpublished results, the full repo is kept internal. Instead, we publish zip archives of the source code for the latest application release. Everything necessary to build the application should be in there.
The Gamma-Ray pulsar search (FGRP) code is not public for a number of reasons. One is that it incorporates code that we don't own the copyright to. If you have a reasonable interest in seeing the code, you may ask for special permission and may need to sign an NDA or sth.
BM
BM
AH - thx.
)
AH - thx.
FWIW I was just interested to
)
FWIW I was just interested to found out which dev can be informed about a typo in current application : "End of stcaktrace"
(http://einsteinathome.org/task/425245058)
Gosh, I fixed this typo years
)
Gosh, I fixed this typo years ago at the corresponding place in the GW search ...
The error, however, is pretty weird, as it actually happens in BOINC code (in boinc_sleep()). Looks like this isn't thread-safe ... Then again, a SIGHUP could happen anywhere anytime ...
BM
BM
RE: Gosh, I fixed this typo
)
So worth to be reported to the boinc dev mailing list ? Or is that something which can be wotk around in the E@H code ?
RE: Gosh, I fixed this typo
)
Are you sure? I can't find 'stcaktrace' anywhere in a current BOINC git checkout - nor even 'stacktrace' ;)
I was thinking it would be better to make a dev list report about 'not threadsafe', and tack the typo report on to that. But it looks like we'll have to leave both to Bernd.
RE: The error, however, is
)
Hhm, as Richard already told you, this is not an issue in the BOINC source code itself (I do follow that git tree, FWIW I'm running 7.3.10 here).
As I already said - it is not really easy to have a quick view at the source code of E@H IMO to quickly find/identify such things. For other FOSS projects like wireshark or the linux kernel I experienced a much lower barrier to make the first steps and therefore to participate in the development of a open source project.
RE: RE: Gosh, I fixed
)
Well, the typo is in the signal handler, which is our code, like in the GW app code.
Well, "non-threadsafe" was just a first thought. However, SIGHUP is a signal that could catch any application anytime, and it's not issued by a system call used in boinc_sleep() like gettimeofday(). At second glance it looks like the origin of this problem lies outside the application, i.e. neither in our nor in BOINC's API code.
BM
BM