2009/6/23 William Stein <wst...@gmail.com>: > > On Tue, Jun 23, 2009 at 10:36 AM, John Cremona<john.crem...@gmail.com> wrote: >> >> That sounds quite sensible to me. > > What is "that"? It sounds below like you're basically arguing for > what we currently do.
I wasn't very clear, sorry. I thought that Nicolas made some good points but did not go through them one by one as I was leaving for a meeting... > > Regarding what we currently do, this is not something that is > "convention emerging" or "standardization attempt". It's something > that Michael Abshoff standardized on probably 8-10 months ago, and as > far as I know strongly required (at least, I remember getting a *lot* > of complaints from him and others when I didn't use that convention). > I personally had another convention, which was: > reponame-tracnumber-description_of_it.patch > > Having been convinced by Michael and others many many months ago to > switch to trac_tracnumber-description_of_it.patch, even though I > didn't personally find it optimal, I now think we should continue with > this standard, since it does seem to work well in practice. > >> Sometimes I make a patch before >> opening a ticket, so the patch name does not have the ticket number on >> it (e.g. #6386 opened yesterday). But it would not be a bad thing if >> I had opened the ticket first (to indicate that I was working on it) >> so that I would have had the number available when creating the patch. > > Indeed. The current trac naming convention really strongly encouraged > you to "do the right thing", which is to always open a trac ticket for > whatever you're working on. > So that's an argument for having a ticket number as part of every patch which his uploaded to trac. > > I think the following is a counterexample to "The trac_ prefix does > not bring any useful information." > >> I like to be able to clear out all the patches which lie around on >> various computers. I know that any patch with a trac number on it >> will exist on trac so can be deleted; anything without a number is >> more likely to be some work inprogress which I have saved from one >> Sage build to carry on with on another, so I tend not to delete those. >> Not quite, since having the ticket number anywhere in the patch name would do that. Nicolas wanted it later in the name, replacing the "trac_" prefix with something more descriptive. One thing I did not like about Nicolas's scheme was that trac names become rather long. Then we need conventions for followup patches on tickets (reviewer's patches and the like). And a convention for whether the reviewer's patch replaces the original (something all too easy to happen by mistake when using MQs, at least for me) or is to be applied after the original. The latter makes it easier for the original patcher to see what the reviewer wants changing; the former makes it easier for others to apply the patch(es). In many cases the reviewer does not make any new patches, just suggests what might or should be changed (more like a referee's report on an academic paper). In other cases there is more of a dialogue between original patcher and reviewer, ending up with a collaborative effort. I think that can be quite productive. John >> John > > "Starting the patch name with the ticket number defeats tab > completion when sorting through a large number of patches, > typically in a mercurial queue. It is a life saver for me to be > able to do hg qpop categories-fra<tab>" > > This could perhaps be solved via technical methods, e.g., some > improvement to how the tab completion works. > > William > >> 2009/6/23 Nicolas M. Thiery <nicolas.thi...@u-psud.fr>: >>> >>> Dear all, >>> >>> Apparently, there is a convention emerging to name systematically all >>> patches on trac as trac_####_description.patch. I very much value this >>> standardization attempt, especially in a period where things are >>> getting automatized. We need it! In particular, I find it very useful >>> to include the trac ticket number. On the other hand, let me argue >>> about some inconveniences of the current naming scheme: >>> >>> - The trac_ prefix does not bring any useful information. >>> >>> - Starting the patch name with the ticket number defeats tab >>> completion when sorting through a large number of patches, >>> typically in a mercurial queue. It is a life saver for me to be >>> able to do hg qpop categories-fra<tab> >>> >>> Thoughs? >>> >>> For the record, here is the naming scheme we use in Sage-Combinat: >>> >>> the_theme-the_description-ticket_number-author_initials.patch >>> >>> example: >>> >>> categories-freemodule-6136-nt.patch >>> >>> Best, >>> Nicolas >>> -- >>> Nicolas M. Thiéry "Isil" <nthi...@users.sf.net> >>> http://Nicolas.Thiery.name/ >>> >>> > >>> >> >> > >> > > > > -- > William Stein > Associate Professor of Mathematics > University of Washington > http://wstein.org > > > > --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
