El Martes, 6 de Octubre de 2009 22:45:20 Austin, Alex escribió: > > -----Original Message----- > > From: openocd-development-boun...@lists.berlios.de [mailto:openocd- > > development-boun...@lists.berlios.de] On Behalf Of Grant Edwards > > Sent: Tuesday, October 06, 2009 12:52 PM > > To: openocd-development@lists.berlios.de > > Subject: Re: [Openocd-development] Moving to git > > > > On 2009-10-06, Ra?l S?nchez Siles <rsanch...@infoglobal.es> wrote: > > >> !!! And if anyone objects to GIT, please speak up ASAP !!! > > > > > > I think mercurial [0] could be a good option. It's a well > > > known distributed VCS, maybe not as spread or known as git, > > > but it's fulfilling most if not all general users aspirations. > > > I find it generally speaking easier to use than git and the > > > move from those used to SVN should be less painful. > > > > This exact same discuss is taking place in a different > > open-source project in which I participate, and several people > > there have also said that they thought mercurial would be an > > easier transition than git. > > > > I would give a few points to mercurial for being written in > > Python. I would expect it to be a lot more stable than > > something written in C. However, my opinion shouldn't be > > weighted very heavily since I'm just a user when it comes to > > openocd. > > I also generally prefer python. However, Python apps are really only > more stable than C apps insofar as segfaults go. I would dare say git > gets much more solid testing, as it's used by the Linux kernel, > where just about every new feature will be tried and beaten on. Only > one unstable release, ever (so far) in the history of git, has had > data consistency problems, and the warning went out on the mailing list > within a few minutes. That was years ago, and hasn't happened again since. > > If you watch Linus' Google Tech Talk on Git, he makes several points > about workflows that are only possible with the performance you get > from the C-based core of git. He discusses Mercurial as interesting, > but it's Python base will cause it to perform worse for many operations. > See http://www.whygitisbetterthanx.com for some performance/feature > comparisons including with Mercurial. >
I said I didn't want to start a flamewar and provided there has been, at least, a slight interest on my messages, I'll try to clear up some point and leave the thread alone. I don't think deciding what VCS to use given the programming language used or only performance benchmarks would be a poor decision. Both focus would help to make your mind what is best, but there should be a list of points to pay attention to, not just those. Needless to say that aspects as usability, support or migration effort from another VCS. Regarding the benchmarking, python core, i.e.: key low-level operations, are written in C. Benchmarking VCS is not that as an easy task. Main problem is comparing equivalent operations, for instance commit in git does not the same thing as in mercurial. Something more measurable would be comparing add+commit for each VCS. That benchmark is not taking into account wire protocol, very important for operations like clone, where I think Mercurial is more efficient. Also there's something weird in branch comparison, I didn't feel that branch on mercurial is slower compared to git. I've asked some people which have more experience with mercurial (and git), and we got to the conclusion that git and mercurial nowadays performs almost the same, differences in regular operations are so small that wouldn't get noticed. There can be, by the way, pathological cases where git will outperform mercurial and some others where the opposite will happen. Most of the time involved in VCS operations is checking for modification in files, where filesystem is to pay attention to. In this aspect either git or mercurial, uses the same underlying filesystem primitives, AFAIK, that's why results are so similar. Anyway, same as already Grant said, I'm just a user, and I've seen some developers are already using git. Maybe it's not convenient to make them move to another VCS at this stage. Regards, -- Raúl Sánchez Siles Departamento de Montaje INFOGLOBAL, S. A. * C/ Virgilio, 2. Ciudad de la Imagen. 28223 Pozuelo de Alarcón (Madrid), España * T: +34 91 506 40 00 * F: +34 91 506 40 01 _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
