packages requiring "httpd" as opposed to requiring "webserver"?
Robert P. J. Day
rpjday at crashcourse.ca
Thu Feb 20 13:43:47 UTC 2014
currently playing with various features of rpm/yum/repoquery, and a
general philosophy question. i can see that, when i query some
packages, they officially "provide" (among other things) a generic
service name. for instance, on my f20 system, i can see that the httpd
package provides the facility of a "webserver":
$ rpm -q --provides httpd
webserver
mod_dav = 2.4.6-6.fc20
httpd-suexec = 2.4.6-6.fc20
httpd-mmn = 20120211
httpd-mmn = 20120211-x86-64
httpd = 2.4.6-6.fc20
httpd(x86-64) = 2.4.6-6.fc20
$
i can use "repoquery" to check on uninstalled packages to get the
same information, say for the "nginx" web server:
$ repoquery -q --provides nginx
config(nginx) = 1:1.4.4-1.fc20
nginx = 1:1.4.4-1.fc20
nginx(x86-64) = 1:1.4.4-1.fc20
perl(nginx) = 1.4.4
webserver <-- there it is again
$
so both of those packages provide being a webserver so, ostensibly,
another package can simply say it has a dependency of a webserver
without having to be specific, correct?
but if i check, say, the "cacti" package, i can see that it
*specifically* wants httpd:
$ rpm -qR cacti
/bin/sh
/bin/sh
/bin/sh
/bin/sh
/sbin/service
/usr/bin/perl
/usr/bin/php
/usr/sbin/useradd
config(cacti) = 0.8.8b-3.fc20
httpd
... snip ...
is that considered good package design? it may be that there are
some packages that absolutely need some webserver feature that is
provided only by httpd, but is it also possible that some packages are
being unnecessarily restrictive? just trying to understand the
packaging philosophy here.
rday
--
========================================================================
Robert P. J. Day Ottawa, Ontario, CANADA
http://crashcourse.ca
Twitter: http://twitter.com/rpjday
LinkedIn: http://ca.linkedin.com/in/rpjday
========================================================================
More information about the users
mailing list