Re: tlug: Mozilla, uh, M15

[shimpei@example.com]

Stephen Turnbull wrote:
>     Shimpei> Right? Right?...Oh, I see. They *don't* provide patches
>     Shimpei> between milestones.  The bastards.
> 
> Why do that?  Every project I know of that has changed patch policy
> has changed in the direction of only providing patches between
> milestones.  It is too easy to screw up, unless you go to the trouble
> of doing fresh CVS checkouts of both revisions.  It also means that
> you have to go to the trouble of quality control on the patches.

I guess I wasn't clear. That's exactly what I'm asking for--just patches
from M15 to M16, or M16 to M17. People who want to try the weekly snapshots
should be subscribing to the CVS tree, as you say, but that's a lot of trouble 
to ask for for people who are only interested in compiling the milestone
releases. 

>     Shimpei> CVS is a pretty server-unfriendly way of disseminating
>     Shimpei> inter-release patches, but if mozilla.org wants it that
>     Shimpei> way, that's their problem.
> 
> Do we distribute patches for the benefit of servers?  I thought we did
> it for developers.  ;-)

Patches for the development versions, yes. Release versions, no.
You aren't seriously suggesting that the Linux kernel patches are primarily
for developers, are you? While developers do use them, they are there
mainly to keep ftp.kernel.org from getting slashdotted every time there is
a kernel update. (It gets slashdotted anyway, but at least there is a
higher threshold.)

> CVS also makes it possible to back out bad changes in a regular way,
> without the cooperation of the release engineer.

Well...yes, sorta, but I'd be wary of any project that uses CVS as a
substitute for a release engineer. Kind of like using a hammer as a
substitute for a carpenter.

> OTOH, I've been doing approximately monthly updates on XEmacs over a
> (normally) 14.4k connection for a while, and that doesn't bother me
> much, takes 30 minutes sometimes.  Maybe twice as long as with a
> direct connection.  I think that probably much of the time is in the
> server generating patch and updates, not in transmission.  (I know
> that most servers request limiting compression to -z3, so there is
> some server burden---but that's their problem, as you point out.)

That's about what I see when I sync a much smaller project over CVS every
night. Of course, the time spent generating patches/updates versus time spent 
transmitting is irrelevant for dialup access customers because we get
billed for both. 

I guess my point is this: CVS is great if you're a developer--much better
than hand-applying a pile of patches. I keep up with a couple of free
software projects this way and I certainly would not bother to do so if it 
involved downloading and patching things by hand every night. 

However, it's not so great if you are just a user trying to compile from 
source--aside from the extra bandwidth involved, you're never sure which
tag on which branch is the "best" release unless you actually follow the 
developers' mailing list. Not to mention the fact that you have to learn
CVS, which is a bit much to ask for from non-developers who just want
to use the latest general release.

Shimpei.


-----------------------------------------------------------------------
Next Technical Meeting: July 8 (Sat)  13:30  Place: LinuxProbe Hall
Next Nomikai meeting: August 18 (Fri) 19:00  Place: TBD
-----------------------------------------------------------------------
more info: http://www.tlug.gr.jp        Sponsor: Global Online Japan