Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 7 Oct 2002 19:56:15 -0700
From:      Steve Kargl <sgk@troutmask.apl.washington.edu>
To:        "Greg 'groggy' Lehey" <grog@FreeBSD.org>
Cc:        "M. Warner Losh" <imp@bsdimp.com>, FreeBSD-current@FreeBSD.org
Subject:   Re: Removing old binaries (was: Do we still need portmap(8)?)
Message-ID:  <20021008025615.GB34673@troutmask.apl.washington.edu>
In-Reply-To: <20021008020141.GP57557@wantadilla.lemis.com>
References:  <20021007063250.GF14070@wantadilla.lemis.com> <20021007.112056.119814448.imp@bsdimp.com> <20021007234610.GT14070@wantadilla.lemis.com> <20021008004442.GA34414@troutmask.apl.washington.edu> <20021008010442.GD57557@wantadilla.lemis.com> <20021008014635.GA34673@troutmask.apl.washington.edu> <20021008020141.GP57557@wantadilla.lemis.com>

next in thread | previous in thread | raw e-mail | index | archive | help
I wrote:
>>
>> I understand what the topic is.  I don't understand your comment, "I'd
>> be inclined just to remove all files in those directories which are older
>> than some file in the build tree--*after* a successful
>> installation."
> 
> Ah, sorry, that might bear more explanation.
> 
> > "install -C" doesn't change the timestamp, so you'll have tons of
> > files that are older than "some file in the build tree".
> 
> What does the last access timestamp look like after install -C?

I'm not sure at this point in time (I haven't looked at the source
in a year or so).  I know "install -C" preserves the last modified
and probably the created time.  

> 
> > You don't blindly want to remove files and I doubt you want
> > mergemaster to list possibly hundreds of files as removal
> > candidates.  So, yes, "install -C" confuses the issue :-)
> 
> Indeed.  What good reason do we have to use it on these directories?

On slow hardware, "install -C" can significantly speed up a
"make installworld", because the installworld doesn't have to
copy any files from /usr/obj to some destination.

I'm beginning to think a mtree.obselete is the way to go.
Each committer, who deletes something from the base system,
should be required to update mtree.obselete.  I think we
should also add a "make purifyworld" or a new mergemaster
option should invoke the mtree.obselete to clean the tree.

I normally use touch(1) and find(1) to eliminate cruft.

-- 
Steve

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20021008025615.GB34673>