Dean Roehrich <Dean.Roehrich at sun.com> writes: > On Wed, Jan 30, 2008 at 12:07:10PM -0500, Richard Lowe wrote: >> James Carlson <james.d.carlson at Sun.COM> writes: >> >> > Dean Roehrich writes: >> >> On Wed, Jan 30, 2008 at 11:09:36AM -0500, Richard Lowe wrote: >> >> > I asked this at the time, and while one person did say something much >> >> > like the above, everyone else's view was that it was the common term >> >> > in use. I'd say if you feel strongly enough that it should be >> >> > changed, come up with alternate but equally understandable/common >> >> > wording and file a bug. >> >> >> >> In patch(1) terms they're called "rejected hunks", or simply "rejects". >> > >> > I don't think we're talking about the same things. "Merge turds," in >> > gatekeeper parlance, are instances where the user has done some >> > intermediate merge step in his workspace, but then hasn't cleaned up >> > after himself by collapsing the SCCS delta or hg changeset. > > Thanks. > > >> > If the user puts back the file, there are no unmerged sections in the >> > body of the file (which are what you'd expect if there were any >> > "rejected hunks"), but the gate's revision history will show swilly >> > entries where the user was hacking around in his own workspace. Over >> > time, this makes the gate's history illegible. >> > >> >> How about another cdm question: What will be the equivalent of "wx >> >> putback" >> >> for internal ON/NV developers? Or is this simply a naked "hg commit"? >> > >> > Yep. >> >> Nope. >> >> Think about it. >> >> "hg commit" would be the equivalent of a wx delget. >> "hg push" would be the equivalent of putback. > > Okay, so 'hg commit' will be naked, and won't have a cdm wrapper to make sure > the commit message is in the correct form (PSARC case, or CR number and > synopsis, on first line). Is this covered by one of the commit hooks?
We really don't want to do that on commit, I don't think. > And 'hg push' would of course be naked. There would have to be a mechanism > for registering one's public ssh key, I suppose? > > Will one of the hg commit or push hooks invoke cdm for sanity checks? But > even with wx apparently the sanity checks are not bullet-proof (see > 58030e1dc0c2). We intend to have the hooks on the gate refuse the more outrages classes of error (bad comments, merges, etc). I'd like to have cdm do similar (and perhaps very slightly more) on the client side of push, too, but we currently don't. wx-like nagging on add/checkin is something we'd also hoped to do (though not with comments on checkin, as above). If people have a specific list of what they feel needs to exist when, it'd help a great deal if they'd post it here, and/or file/update related bugs with us. -- Rich
