Hey
Tomorrow, on March 20 2010, it is exactly 12 years since the first release of
curl. It is about time to freshen up a few things in this project.
What am I talking about? Version control and Bug tracking mostly. I'll
elaborate below. This is what I feel we should do to inject some freshness
into the project and get rid of some of the rough edges that we still have.
I'm listening to your feedback. Am I wrong or am I right?
Version control
---------------
CVS has served as well for many years, but time has come to finally get rid
of it and enter the age of good tools. I'm using proper tools more and more
at work and in other projects so the quirks of CVS has made me growingly
uncomfortable over time.
The only sane way forward as I see things is to go git. With git we get much
better handling of patches (git am), we get much better handling of original
author (git am and git commit --author etc) and it allows for a lot more
fancy stuff that might be handy at times (like offline work and branching
and...).
CVS is easy and git is rather hard to use. Will it cause some annoyances and
mistakes? Yes, but I still think it is worth it for us to cross that bridge
in the long run.
I'm experimenting with CVS conversion stuff right now so that we get the
full history properly (well, since 1999 which is the furthest we have) and I
think I'll then proceed and host the repo on github.
A little outstanding question is how to deal with c-ares which so far has
been living its life "embedded" in the libcurl tree and I figure perhaps now
we've reached a point where we split this tight relationship and turn them
into two separate repos. I'll also bring this up for discussion on the c-ares
list.
Bug tracking
------------
We abandoned Sourceforge for most services years ago (since it was slow and
unreliable), but we hung on to the bug tracker they host for us. That bug
tracker is annoying (to many people) and for some reason it is always
notoriously slow. The time has come to get away from SF for that final piece
as well.
We don't have any particularly weird or strange requirements on a bug
tracker, and I've been wanting someone to offer us to host a trac or
something but it hasn't happened yet so I've decided I'll host it on the
regular curl web server (owned and run by Haxx). My plan is to import the SF
entries to trac and disable the SF one.
--
/ daniel.haxx.se
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html
