Is urlgrabber a premature optimization?

[Aaron Bentley]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi all,

AIUI, the key advantage of urlgrabber over urllib is that it supports
keepalive, which makes batch downloading somewhat faster.  But it
doesn't support pipelining or parallel downloading, so for batch
downloading, it's not really adequate.

If we switch to effbot.org or twisted, urlgrabber will not provide any
advantages, because it will only download one file per session.

Currently, support for urlgrabber is on by default.  When urlgrabber is
not available, bzr cannot use RemoteBranches.  It has been suggested
that bzr should fall back to urllib if urlgrabber is not available.

However, if urlgrabber is really a premature optimization, perhaps we
should simply disable support for urlgrabber completely.  Insert
standard justifications about maintainability, testing, dependencies,
etc here.

Urlgrabber isn't current available as part of Debian or Ubuntu, which is
blocking bzr from entering those distros.  Rob Weir is willing to
package urllib, but if it's not needed, we can save him the trouble.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCuxwE0F+nu1YWqI0RAlJEAJ9sEWokIspx9DbFcPVeb2JYiaJ/HQCcDJ3g
BSGerBzht+sIBwyQ3qRRqHI=
=bI2v
-----END PGP SIGNATURE-----