>>>>> Richard Stallman <address@hidden> writes:
> I think we should state that as a lower bound, a minimum period to wait, not
> as a default. To have a default waiting period before deleting obsolete
> features would lead us to act rigidly, rather than considering in each case
> what is best for the users.
I'm quite OK with being flexible about when we feel it's time to delete old
code.
To summarize the final position: Once code has been in "obsolete" for a
complete release cycle, removal may be recommended by those who feel it's time
for the code to go. Once it's truly gone, any bugs against it and remaining
documentation can follow.
I think that summary has one key difference to what we previously discussed, and I think that difference is problematic.
My goal in starting this thread was to find a way for emacs to be able to remove functionality. But the proposal, as John puts it, would in effect keep code around forever. Someone will have to, at the right time, make an argument that the code should be removed, and be willing to advocate for it. This is too high a burden for code that's been in obsolete for years. I think the burden should be on people to advocate for it's continued existence.
Specifically, we can, at the start of a major release, put out a list of all obsolete functionality that the maintainer feels is safe to delete, and ask if anyone has objections to deleting them.
Making the default to delete instead of keep will have a real difference on how much code actually gets removed.
--
John Wiegley GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com 60E1 46C4 BD1A 7AC1 4BA2