[Yum] script run idea

Erik Williamson erik at cpsc.ucalgary.ca
Thu May 29 15:44:43 UTC 2003


Your suggestion makes a lot of sense:

> config /path/to/config
> install foo
> update bar
> remove baz
> groupinstall quux
> config /path/to/config2
> install do
> remove re
> update mi
> groupinstall fa
> config /path/to/config3
> install tom
> remove harry
> update john

Then I got thinking of in that case, there'd be a 'meta-config', and 
three other individual configs.  This got me thinking about a yum.conf 
that looks like this (please to forgive simulated XML):

--- warning: this is just a brainstorm ---

<yumconfig>
	<option name="cachedir" value="/path/to/cache">
	<option name="debuglevel" value="2">
	<option name="logfile" value="/path/to/logfile">

	<group name="redhat">
		<repository name="Redhat 9 - base">
			<baseurl location="http://path/to/repo">
			<baseurl location="http://path/to/failover">
			<failover method="roundrobin">
		</repository>
		<repository name="Redhat 9 - updates">
			<baseurl location="http://path/to/repo">
		</repository>
	</group>
	<group name="extras">
		<repository name="FreshRpms - RedHat 9">
			<baseurl location="http://path/to/repo">
		</repository>
	</group>
</yumconfig>

This would get rid of the nice & clean existing syntax though. I don't
know how many people would need/utilize this.  but, the 'template' could

then have the following format, or some such:

<template>
	<group name="redhat">
		<action type="groupinstall">tom</action>
		<action type="remove">harry</action>
		<action type="install">ted</action>
		<action type="update">*</action>
	</group>
	<group name="freshrpms">
		<action type="install">do</action>
		<action type="update">*</action>
	</group>
</template>

obfuscated...  but just a thought.

-----Original Message-----
From: yum-admin at lists.dulug.duke.edu
[mailto:yum-admin at lists.dulug.duke.edu] On Behalf Of seth vidal
Sent: Thursday, May 29, 2003 8:51 AM
To: Erik Williamson
Cc: yum at linux.duke.edu
Subject: RE: [Yum] script run idea


> 
> I like your idea here; reason being that with my local system, each
host
> here uses yum twice - one config for the redhat release and it's
updates
> (using "yum -c http://blarg update"), and one for 'extras', like
select
> freshrpm packages, and whatever else people ask for once I do the
> rollout (using " yum -c http://foo install \* ").  Looks like this
could
> take care of that.

but how do I parse those for the purposes of one transaction?

an example:

install foo -c config1
install bar -c config2
install baz -c config3

now - if I want those handled as one transaction, I'm going to have to
read config1, figure out where to get foo and its deps, then read
config2 and do the same for bar

but how do I finish the process?

ie: once I know all the steps I will need I'm not done with the config
information. I need the config info to know where files are.

But if you're saying you handle everything up to a -c in one
transaction, then restart for a second config file, that's a different
story.

so something more like this:

config /path/to/config
install foo
update bar
remove baz
groupinstall quux
config /path/to/config2
install do
remove re
update mi
groupinstall fa
config /path/to/config3
install tom
remove harry
update john

that would be more doable - in 3 separate transactions
-sv





> 
> I agree with Troy about the syntax; the command-line syntax is great.
> 
> e.
> 

_______________________________________________
Yum mailing list
Yum at lists.dulug.duke.edu
https://lists.dulug.duke.edu/mailman/listinfo/yum




More information about the Yum mailing list