From a release management standpoint, the new CVS log accumulator is a
big step backward. It used to be possible to get a decent idea of
who's doing what by just looking in the MUA summary window, but no
1. I miss the friendly "CVS Monitor" From header. (whi-i-i-i-ne :)
2. The new subject line is essentially useless. Specifically, the
committer is missing, and much of the time the tiny amount of
repository information is far too broad to be helpful.
3. This is especially annoying if you don't have working broadband (I
pay for it, but I don't get it half the time) because it means you
need to download megapatches just to get committer information.
4. The stable RE is going to have to pay close attention to *both*
-patches and -cvs if and when we go to commit-without-submit on the
assumption that the commit diffs are good enough. At present, he can
ignore -cvs if he chooses. (Not my problem at the moment, but it
certainly would be a consideration if I'm a candidate for gamma RE or
stable RE in the future.)
5. The commit diffs are not an accurate record for review, and if
iterated cannot be applied to variant workspaces, since they compress
the patch with -w. Even if the patch applies, a diff against the
mainline will be spurious. I know that CVS branches aren't terribly
useful, but this will make them even less so.
The rest of the stuff I can live with, but I really want committer
id in the subject, and I think it's quite desirable to give full
patches, bandwidth be damned.
Other suggestions: I think that it would be nice if the first line of
the log would be a 40- or 50-character "theme statement" that could be
picked up. Something like
Subject: [Commit by ben] 21.5 - rot13 all identifiers for readability
where everything after the  would be from the committer's log. The
"21.5" is a pseudo-module, more or less arbitrarily chosen by the
committer, as is the short description. Other examples would be
unbundled packages such as "xemacs-devel", projects such as "KKCC",
and functional divisions like "redisplay".
Bonus points for snarfing message IDs from the log and putting them in
a X-Cross-Reference header (at least in my own usage, there is no
guarantee that multiple message ids will be threaded, so References is
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.