Source Code Control
in [Embedded]
|
From: Darin Johnson on 7 Nov 2006 16:46 Steve at fivetrees wrote: > Atomic commits also mean that the entire commit operation runs, or none of > it. In other words, something like a network glitch won't wind up with a > broken repository. This is a Perforce feature also, which is pretty nice. The only real drawback I have with Perforce is that the GUI doesn't cooperate well with the Cygwin under Windows command line (but most SCC tools would have difficultywith that sort of thing). I used to use CVS not too long ago, which was ok, but I really really hated its automatic merge method, which it would occasionally screw up (despite the authors' insistence that this never happens unless project members fail to communicate effectively). I alway preferred the manual three-way merge: your changes vs common ancestor in one pane, their changes vs ancestor in another, and the merged stuff in the center pane. CVS gives you no convenient way to get this information since it's all done deep down using a "diff3" reimplementation. Perforce does have a nice merge tool though (choice of manual or automatic). If Subversion has gotten around this merge headache, then I may consider it again in the future. -- Darin Johnson
From: Steve at fivetrees on 7 Nov 2006 16:51 "Darin Johnson" <darin(a)usa.net> wrote in message news:1162935981.699191.232800(a)b28g2000cwb.googlegroups.com... > > I used to use CVS not too long ago, which was ok, but I really really > hated its automatic merge method, which it would occasionally > screw up (despite the authors' insistence that this never happens > unless project members fail to communicate effectively). > I alway preferred the manual three-way merge: your changes vs > common ancestor in one pane, their changes vs ancestor in > another, and the merged stuff in the center pane. CVS gives > you no convenient way to get this information since it's all > done deep down using a "diff3" reimplementation. Perforce > does have a nice merge tool though (choice of manual or automatic). > > If Subversion has gotten around this merge headache, then > I may consider it again in the future. I think it has. The merge tool looks pretty good. I've seen it only briefly, but a colleague was very pleased with it. Steve http://www.fivetrees.com
From: Pete Fenelon on 7 Nov 2006 16:48 Henrik [6650] <henrik6650(a)ofir.dk> wrote: > > I sit here wondering if all this Anti VSS propaganda out there is due to a > general disliking of Microsoft, or does anybody actually have any bad > experiences with Visual Sourcesafe? Many. In a PPOE we had about six databases, and every one of them was "corrupt" in ways that the MS analyze utility could not diagnose or repair. Checkins and checkouts would randomly fail. It was impossible to script earlier versions of VSS because it was tied to one INI file linking it to the database. Performance became glacial as two files were created for every checkin - the VSS filesystem became fragmented and *s...l...o....o....w....*. Sharing and branching became phenomenally expensive, and eventually unreliable. And migrating from VSS5 to VSS6 was basically impossible, given that the VSS5 databases were "corrupt". I don't trust it, and I understand that MS don't use it either. pete -- pete(a)fenelon.com "he just stuck to buying beer and pointing at other stuff"
From: Pete Fenelon on 7 Nov 2006 16:50 Grant Edwards <grante(a)visi.com> wrote: > Sure. At one employer we lost files and ended up with corrupt > databases on multiple occasions. Since we weren't running the > latest version, no support was available from the vendor. The > portion of that project that I worked on personally was under > RCS -- which never a single problem. > > At another employer, the VSS database repeated got "corrupted". > Each time it happened, the whole thing had to be shut down and > fixed in a weekend-long process. A number of files and > revisions were lost. Been there, done that, worked the late nights trying to unscrew the screwed pooch. > > One of those companies has ditched VSS completely, and the > other is in the process of changing over to SVN. I took all my projects off VSS and into CVS and eventually Subversion. Life became much, much easier. pete -- pete(a)fenelon.com "he just stuck to buying beer and pointing at other stuff"
From: Steve at fivetrees on 7 Nov 2006 16:56
"Grant Edwards" <grante(a)visi.com> wrote in message news:12l1ue4fbahft00(a)corp.supernews.com... > On 2006-11-07, Henrik [6650] <henrik6650(a)ofir.dk> wrote: > >> I sit here wondering if all this Anti VSS propaganda out there is due to >> a >> general disliking of Microsoft, or does anybody actually have any bad >> experiences with Visual Sourcesafe? > > Sure. At one employer we lost files and ended up with corrupt > databases on multiple occasions. Since we weren't running the > latest version, no support was available from the vendor. The > portion of that project that I worked on personally was under > RCS -- which never a single problem. > > At another employer, the VSS database repeated got "corrupted". > Each time it happened, the whole thing had to be shut down and > fixed in a weekend-long process. A number of files and > revisions were lost. > > One of those companies has ditched VSS completely, and the > other is in the process of changing over to SVN. > > There's a huge, basic flaw in VSS's design. It's a single, > monolythic, database file with no centralized server or > control. All of the clients bang away individually at the > database file. If one client glitches the whole thing crumples > to the ground like a house of cards. Yep. And the file structure (e.g. files named "aaaaaaaaay") and format (effectively unreadable) is odd. We also have experienced database corruption and missing source files. One chunk of code, fortunately not too vital, went missing for years... curiously, when I converted the database to Subversion, it showed up again... Steve http://www.fivetrees.com |