On Tue, May 28, 2013 at 10:45:25PM -0400, Ryan Kavanagh wrote: > On Tue, May 28, 2013 at 05:08:06PM -0700, Steve Langasek wrote: > > There have been two responses to your proposal so far, neither of > > which particularly looks to be in favor of your plan. I don't think > > it's reasonable to proceed with a mass-bug filing on over 800 packages > > as a first step, certainly not after such a short comment period. > As for the package count, where did you find the additional 700+ > packages? If I'm missing something, please let me know. Of matches in > the search[0] Paul Wise linked to, many matches are from the same > package, or more prominently (as with e.g., festival, gcc, geiser, > id-utils, gcl, etc.), found in the upstream build system but never > called at build time (one can deduce this from the fact that they don't > build-depend on texi2html). As I stated in my initial email, there are > at most 96 packages which would require some form of change, and these > are the packages that either build-depend (94) or depend (2) on > texi2html. Oops, sorry, I misread Paul's mail. I see now that he wrote that there were "877 pages worth [of matches] in Debian sid", not that there were 877 *packages*. So 100 packages does seem much more reasonable to me for an MBF, yes. > > Why are you not proposing to provide a texi2html wrapper from the > > makeinfo package which translates the arguments as described on > > <http://wiki.debian.org/Texi2htmlTransition>, and have makeinfo > > Provide: texi2html? > If by this, you mean remove the texi2html package from the archive after > introducing the texi2html wrapper for makeinfo, and hope people would > then transition after being warned by lintian, then it would fail to > address Sébastien's concerns. That concern being that makeinfo is not a drop-in replacement for texi2html functionality, and therefore this would cause regressions - ok. > If you mean something else, could please explain in greater detail what > you meant, and how it would allow me to remove the texi2html package > much more quickly from the archive than with my current plan? It is still *potentially* faster, in the sense that in both cases, texi2html can't really be removed until makeinfo is a feature-complete replacement, but in one case we have to wait for all the other reverse-deps to also be updated whereas in the other case we *only* need updates on the makeinfo side (@math support + wrapper) to unblock the texi2html removal. I guess it's a question of whether we think @math support is going to be on the critical path for this transition. > > This could be coupled with a lintian warning, or other soft means of > > encouraging maintainers to do the transition. > Given the above, I think the best approach, assuming there's no further > support of the MBF over the next week, is: > * start off with a lintian warning (lintian maintainers willing); > * in 3, 4, or 6 months time, proceed with a mass bug filing against > the remaining packages. > Steve, does this seem reasonable to you? Now that I've read the numbers correctly, I don't mind an MBF here if that's what you prefer. But a lintian warning could probably also be a good approach. Cheers, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. Ubuntu Developer http://www.debian.org/ slangasek@ubuntu.com vorlon@debian.org
Attachment:
signature.asc
Description: Digital signature