On Thursday, April 9, 2015, Magnus Hagander <mag...@hagander.net> wrote:
> On Thu, Apr 9, 2015 at 2:20 PM, Tom Lane <t...@sss.pgh.pa.us > <javascript:_e(%7B%7D,'cvml','t...@sss.pgh.pa.us');>> wrote: > >> Magnus Hagander <mag...@hagander.net >> <javascript:_e(%7B%7D,'cvml','mag...@hagander.net');>> writes: >> > On Apr 9, 2015 2:20 AM, "Robert Haas" <robertmh...@gmail.com >> <javascript:_e(%7B%7D,'cvml','robertmh...@gmail.com');>> wrote: >> >> +1. >> >> > Is that at +1 for naming it moved, or for not having it? :-) >> >> > I can definitely go with moved. Buy I would like to keep it - the reason >> > for having it in the first place is to make the history of the patch >> follow >> > along when it goes to the next cf. If we don't have the move option, I >> > think it's likely that we'll be back to the same patch having multiple >> > completely unrelated entries in different cfs. >> >> The problem with the whole thing is that you're asking the person doing >> the "returned" marking to guess whether the patch will be resubmitted in >> a future CF. >> >> The right workflow here, IMO, is that a patch should be marked returned or >> rejected, full stop; and then when/if the author submits a new version for >> a future CF, there should be a way *at that time* to re-link the email >> thread into that future CF. >> > > If we just link the email thread, that would mean we loose all those > precious annotations we just added support for. Is that really what you > meant? We also loose all history of a patch, and can't see that a previous > version existed in a previous commitfest, without manually checking each > and every one. But if that's a history we don't *want*, that's of course > doable, but it seems wrong to me? > > I'm not necessarily saying that what we have now is right, but just giving > up on the history completely doesn't seem like a very good workflow to me. > > We could always tell those people to "go back and find your old patch and > re-open it", but in fairness, are people likely to actually do that? > > > "Moved" is really only applicable, I think, for cases where we punt a >> patch to the next CF for lack of time. >> > > Well, that's basically what "returned with feedback" is now, so I guess > that one should just be renamed in that case. And we add a new "returned > with feedback" that closes out the patch and doesn't move it anywhere. > Which is pretty similar to the suggestion earlier in this thread except it > also swaps the two names. > > Can we create a "fake" CF time period into which all of these "waiting on author" entries can be placed and readily browsed/found instead of leaving them in whatever CF they happened to stall in? David J.