On 7/11/07, Ben Coburn <[EMAIL PROTECTED]> wrote:
Obviously each project has it's own circumstances, so many projects will use different languages, revision control systems, editors, etc., etc., etc.. If Mercurial works for you, that's great.
The fact that this problem occurs rather often suggests that there are serious issues with darcs as it stands. It's happened to me at work (where I now use git because of it), it's now happened to Zooko's coworker, I know that Simon Peyton-Jones had issues with Darcs not a couple of months ago. I've been on this list for some time, and it seems that someone has a critical issue once every other month (where I define "critical" as "I'm not using darcs anymore"). The problem is, when darcs has a problem, it is damn near impossible to replicate. The problem I personally had with darcs has still to be replicated on my end. It's almost as if it failed due exclusively to a cosmic ray. Alas, because the data was strongly proprietary to the company I work for, I couldn't just send the contents of the repos as they existed. This is the conundrum. darcs seems like it's perfect when working on most projects, but the one time you use it (with full confidence) on a proprietary project, it'll inevitably screw up in such a manner that you can't recover, AND you can't send enough information to debug with, because the data is under NDA. That being the case, I also use darcs for local stuff and upload my website with rsync. However, this was due out of necessity, since the website administrator was never able to get darcs working on his box (I don't know why, and wasn't about to debug the situation for him. Besides, I'd rather my website be read-only to everyone anyway!).
For comparison, I've been contributing to the DokuWiki project ( http://wiki.splitbrain.org/wiki:dokuwiki ) for a while and have heard very few complaints about darcs from developers over there.
Complaints _are_ few. The problem is, when someone complains on this list, it's usually because darcs just nixed their project's repository somehow.
I might also note that I ran 'darcs get -v http://allmydata.org/ source/tahoe/trunk'. It was quite verbose, and finished in ~1 minute on ordinary hardware with an ordinary DSL connection. Interactive commands might be slower... but http based gets and pulls seem fast enough to me.
You missed the point where he said that he's rsyncing a repo up to the site now. You _will not_ see a problem with this approach, since you're seeing a clone of the developer's local repo, which is known good (or else, he'd never upload it to the site to begin with). It's the same technique I use on my own site. -- Samuel A. Falvo II _______________________________________________ darcs-devel mailing list darcs-devel@darcs.net http://lists.osuosl.org/mailman/listinfo/darcs-devel
