DragonFly users List (threaded) for 2004-11
Re: Retrieving less data on cvsup?
On Mon, 1 Nov 2004 09:46:08 -0500, "George Georgalis"
>On Mon, Nov 01, 2004 at 01:40:33PM +0000, Dark Fiber wrote:
>>On Mon, 1 Nov 2004 00:54:20 +0100, Diego Calleja <diegocg@xxxxxxxxx>
>>>I recently cvsup'ed the FreeBSD ports list. Syncing the ports database has
>>>always been painful with slow connections, even with compression, specially
>>>with freebsd's ports. I've always wondered if there's some way of not
>>>downloading some of the files until I use them, like say
>>>/usr/ports/foo/files/file.patch, which I'm not going to use until I install
>>>the program, if I install it. I've diving trough the handbook but I didn't
>>>find anything, or I missed it. It's possible to do something like that?
>>whait i had thought was that each port should be 1 .tgz file,
>>tar+gzip all the files + patches. this would use a boatload less
>>inodes + disk space.
>but that would make searching descriptions difficult... maybe two files?
your right. when I add a port, I usually know what Im adding.
If i do anything I usually grep the INDEX to see where it is
more than anything.
If you do anything binary, that would make cvsup a PITA
i had initially thought a zip file (with the pkg-descr as a
zip comment, and portupdate could just grab the descriptions
of updated ports). but you end up at the same place as
other poster pointed out with same problems.
what really is needed is a replacement system. that
would nicely take everything conceivable and things
not, and work in a nice sane fashion and be favourable
for everyone :)
it shouldnt be that big a deal tho, we are not talking
about package management, just port management. a makefile,
some diff patches, a description etc.
-- stu --
[FAQ] Write Your Own Operating System