improving rpm

Stephen John Smoogen smooge at gmail.com
Wed Jul 12 19:55:28 UTC 2006


On 7/12/06, Nicolas Mailhot <nicolas.mailhot at laposte.net> wrote:
> Le mercredi 12 juillet 2006 à 15:24 -0400, Neal Becker a écrit :
>
> > 1) Don't create needless .rpmsave/.rpmnew if there is no change
>
> Isn't it already supposed to do this ? (though if I look
> at /etc/ld.so.conf, probably not)
>
> > 2) Attempt to merge changes, in the manner of a modern revision control
> > system.
>
> This is a 100%-proof recipe for disaster, even if conf files where all
> in structured format like XML with full grammar available
>

The only way I have seen this 'work' is where the end result config
file is built from 'other' config files. In this case you would have a
default config, an repository patch, and a local patch. The master
ninja configurator would then build its new config by superceeding
default by any repository patches and those would be replaced by
local. This would then would build out the new configuration file.

The problems were you had to know that 'master ninja configurator'
controlled all of its config files and didn't make local changes.. and
you had to write m4 for each type of end-result config file that would
be created. [By 3 insane cs students who spent too long reading the
sendmail source code to find Cthulhu's Code.]

-- 
Stephen J Smoogen.
CSIRT/Linux System Administrator




More information about the devel mailing list