On Sat, Aug 17, 2013 at 11:47 AM, Keith Medcalf <kmedc...@dessus.com> wrote:
> > On 16.08.2013 18:03, Gary Weaver wrote: > > You're right concurrency is not outlandish but from the viewpoint of > > sqlite it was therefore this answer and the link. > > Hold on here. > > What do you mean "it works in OS X" and "not in Ubuntu VM in Travis"? > > The issue is not the concurrency in SQLite, but rather that the Operating > System is broken because the "concurrency" of the SQLite library does *not* > change -- the code is exactly the same -- the change is the OS. You need > to look at what is unsupported or non-functional in the "Ubuntu VM in > Travis" since that is the root of the problem, and has nothing to do with > SQLite itself. > > I appreciate everyone's comments, but please read the entire thread. I was pointed by a few people to pages in which SQLite documentation indicates that to, on its own, does not handle concurrent writes. I've also relayed everything I've learned about the environment here in the thread, and have relayed it to the TravisCI team in the hope that they might be able to track it down or get Blue Box's assistance in tracking it down. SQLite has been very useful to me in testing. Yes, I was only surprised to learn that it was as mature as it is and not meant to handle concurrent writes, but I'm much less surprised now that I know that it relies on the filesystem to handle locks properly and that concurrent writes typically work for non-networked filesystems, which is what I experienced in OS X. I spent time today doing some research on the environment in which the issue occurred, and have asked for additional assistance from the TravisCI team to help track this down, but I'm just doing this for the purpose of helping other SQLite users, because at this point, it is not affecting me much. As I'm sure you all understand, there are a lot of projects having continuous integration testing on Travis using SQLite. I don't think that mentioning the problem or pursuing a fix is a bad thing. > the Operating System is broken I think that is a little harsh. It just might be having some issues with posix locks (maybe- not yet proven). > You need to look at what is unsupported or non-functional in the "Ubuntu VM in Travis" since that is the root of the problem, and has nothing to do with SQLite itself. If you'd read the thread, you'd have seen that I already started that. Please everyone- lay off. Thanks. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
