Re: cvs commit: squid/src cachemgr.c

From: Takahiro Kambe <taca@dont-contact.us>
Date: Wed, 27 Apr 2005 00:13:38 +0900 (JST)

Hi,

In message <Pine.LNX.4.61.0504260645270.26647@localhost.localdomain>
        on Tue, 26 Apr 2005 07:12:26 +0200 (CEST),
        Henrik Nordstrom <hno@squid-cache.org> wrote:
> It's true that it is annoying to some that the patches is updated,
> certainly so on cosmetic fixes such as the one in cachemgr.c, but most
> often the patch updates are quite critical in nature (fixing bugs in the
> published patch) and the old patch should no longer be used or referenced.
> For us it is a feature that this breaks distfiles, as it prevents rebuilds
> using the broken patch.
Completely broken patch which can not be used to patch should be
replaced. But after that, I wish farther changes would be done by
another patch incrementally.

Or I would happy patch files are under some time stamped directories.

        2005042100/squid-2.5.STABLE9-cachemgr_objects.patch
                   ...
        2005042600/squid-2.5.STABLE9-cachemgr_objects.patch
                   ...

So, latest patches are under latest date of directory.

> IMHO package maitainers should only include the patches seen absolutely
> required by your QA of the packaged release. Generally you should wait
> until the next release making both yours and our life easier. For package
> maintainers the patches is mainly provided in case your QA policy does not
> allow updating to the new STABLE release and requires backporting of the
> relevant changes after the new release is published.
But once I choose to using a patch, previous patch needs to be applied
and always fear chosen patch files would be modified suddenly.

(And package maintainer are sensitive "security fix" patches.)

Anyway, this is my little wish.

-- 
Takahiro Kambe <taca@back-street.net>
Received on Tue Apr 26 2005 - 12:00:09 MDT

This archive was generated by hypermail pre-2.1.9 : Sun May 01 2005 - 12:00:06 MDT