On 07.08.2008 18:11, Peter Arrenbrecht wrote: > On Thu, Aug 7, 2008 at 11:12 AM, Adrian Buehlmann <[EMAIL PROTECTED]> wrote: >> I've created a new page on the Mercurial wiki, trying to shade some light >> on what openers do and how they are used in Mercurial: >> >> http://www.selenic.com/mercurial/wiki/index.cgi/Opener > > Thanks Adrian, shall try to look at it tomorrow. I asked on #mercurial > about whether this bug would affect the hardlinks created by `hg > clone` and ThomasAH thought not (as I do). Only files outside of > .hg/store/ hardlinked by operations such as `cp -al`. So unless your > page or my own digging into the code is going to convince me > otherwise, I still think this bug is not that critical on Windows. Not > sure I can do this in time, though, as I'm going on vacation again > Saturday.
I never claimed it to be critical! I just see no point in running the risk of letting it *get* critical, if that damn thing has already been fixed weeks ago! Who knows what dirty tricks new extensions will play? And why let yet another release escape that does not contain the fix for that bug? Again, Mercurial would certainly *not* have released a new release *not* containing the fix for that bug, but THG has done exactly that. It looks like some part in the calling code would have to do modes like "r+" or "rb+" or something similar (did a quick search when Benoit pushed that cset then, I think mq does that). These modes look at first sight like they were read only. But they aren't! They modify the file! It was just quite shocking for me to see such a big botch in such a central part. The code that must be executed in the opener when it is a write access just neglected a whole bunch of cases that actually are write accesses too. Funny thing is, while I was doing my long filename patch odyssey (see mercurial-devel list) I stared at that code there and thought "damn, if that test is wrong, we have a mess with the hardlinks", but then moved on because I thought the Mercurial top shots would have thought that very well, when they hacked that sensitive thing. And bam! A few days later Benoit pushed that cset (note that my patches would not be affected by that because I deliberately don't look into modes -- for good reason). Mercurial and the current extensions seemed to have had some luck until today that nothing vital was hit yet, but facing the theoretical possibility makes my gut feeling to decide to fix that damn thing ASAP. You might think I'm too paranoid about this one, but I have been hit in the past enough times for *not* being paranoid enough. I would also like to point out the even if the THG people claim they do only a GUI thing, their package still contains the core Mercurial and probably 90% of the Windows people which want Mercurial will download and install THG. In light of that, I think it is also a bit unfair to the Mercurial team, to ship that bug in a newly created binary. After all, they have fixed it! They just didn't get around to rubber stamp it with another release tag! (which doesn't mean you can't trust crew-stable). So, you see, we have wasted a lot of time talking about this stupid issue here. Without any need at all. It would have been a simple thing to take crew-stable. Or at least not trying to tell me how foolish and arrogant I am in daring to criticize THG for not doing so! I also don't think you need to look into this code Peter, so please enjoy your holidy. It won't make any difference. Simple recommendation from my side for next time: If you want to do an async release of THG next time and there are important fixes in crew-stable of Mercurial, take crew-stable *or* wait for the next rubber stamped Mercurial (if you still want to have a rubber stamped one). ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Tortoisehg-discuss mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/tortoisehg-discuss
