1 |
As a user I think a nice feature would be to have emerge actually list all the files that were saved as ._cfg... with an indication of if the original was "untouched" or "modified." An example of where this would be very usefull is the pcmcia-cs package. The /etc/pcmcia/config file may be updated to include some new cards and I never touched it because config.opts is where local changes are supposed to reside. Chances are PCMCIA will work fine and I won't even bother to look to see what config files could be updated because I don't have the time to worry about it, however someting may have been changed for a very good reason (possibly change a default for security reasons or some such). Even worse is the baselayout Geert Bevin complains about below. I understand the policy and thing it is sound, but i would still like a --overwrite-untouched that saves the original config file with a ._old prefix and gives a list at the end of config files i overwrote to give me a chance to make shure i didn't do someting i didn't want to (which menas it is only overwriting config files when told by the administrator, and even then doing so very cautiousally and making backups). |
2 |
|
3 |
just my 2 cents :) |
4 |
|
5 |
--Nate |
6 |
|
7 |
On Sat, Dec 08, 2001 at 04:20:26PM -0600, Daniel Robbins wrote: |
8 |
> On Sat, Dec 08, 2001 at 10:19:28PM +0100, Geert Bevin wrote: |
9 |
> > There is one additional feature I'd appreciate very much personally. |
10 |
> > Currently if th file exists, the ._cfg file is created. Imho it would be |
11 |
> > nice if an additional check would be performed first to see if it has |
12 |
> > been modified by the user. If it hasn't, it can be overwritten without |
13 |
> > problems. Now when I'm diffing after installing a new baselayout I spend |
14 |
> > a lot of time on files I've never even touched. |
15 |
> |
16 |
> The way we do it now is a policy decision. We simply do not want config |
17 |
> files to change without the admin knowing about it. Assuming that an |
18 |
> untouched file is "fair game" is still an assumption -- one that could |
19 |
> very easily be wrong. |
20 |
> |
21 |
> Best Regards, |
22 |
> |
23 |
> -- |
24 |
> Daniel Robbins <drobbins@g.o> |
25 |
> Chief Architect/President http://www.gentoo.org |
26 |
> Gentoo Technologies, Inc. |
27 |
> _______________________________________________ |
28 |
> gentoo-dev mailing list |
29 |
> gentoo-dev@g.o |
30 |
> http://lists.gentoo.org/mailman/listinfo/gentoo-dev |