On Wed, Sep 11, 2013 at 8:40 PM, Glenn Fowler <[email protected]> wrote: > > On Wed, 11 Sep 2013 20:34:00 +0200 Irek Szczesniak wrote: >> On Wed, Sep 11, 2013 at 4:44 PM, Glenn Fowler <[email protected]> wrote: >> > >> > On Wed, 11 Sep 2013 16:37:36 +0200 =?KOI8-R?B?z8zYx8Egy9LZ1sHOz9fTy8HR?= >> > wrote: >> >> --047d7b603dccfe0d4104e61c8fa1 >> >> Content-Type: text/plain; charset=ISO-8859-1 >> > >> >> Glenn, this clean up patch is still on hold, right? >> > >> > yes on hold > >> Can I ask why? We're talking about >> astksh20130829_libcmd_at001.diff.txt, right? The patch which replaces >> open() with openat() and sfopen() with sfopenat() and so on? > >> I have trouble understanding this because it should be a low risk patch. > > the problem is it only attacks part of libcmd > fts, which is the basis for all ast directory traversal, needs to grok *at() > too, > and its not as straightforward as the other parts > > why hold off on a partial patch? > because based on past experience the day 2 mail flood will start with > "none of the -R commands work"
1. Roland's mail says: "...I've avoided builtins which use fts for now to avoid crashes between other work like the fts |*at()| work...". The crash part confuses me but the patch itself is pretty clear. 2. The -R commands - assuming you're referring to grep -r, cp -r, mv -r ... - still work. And if I get the patch right they continue to work as long as ((context)?context->pwdfd:AT_FDCWD) points to a directory fd which is equal to the process's current working directory. You don't like patches which do changes in multiple steps, do you? Irek P.S.: I'm not the one who will scream "bloody murder...none of the -R commands work" if its clear that the patch which just went in is the beginning of a patch series. I'd start screaming if someone announces that the patch series is complete (its obvious that Roland's patch is just taking on the easy parts) and then the -R commands don't work. _______________________________________________ ast-developers mailing list [email protected] http://lists.research.att.com/mailman/listinfo/ast-developers
