| 1 |
On Mon, 16 Feb 2004 03:19:17 +0100 |
| 2 |
Marius Mauch <genone@g.o> wrote: |
| 3 |
|
| 4 |
> So therefore I'd be interested if there are any strong reasons |
| 5 |
> to keep this approach or if we could switch to one variable |
| 6 |
> RSYNC_OPTIONS that's predefined in make.globals and users could |
| 7 |
> redefine it in make.conf (of course transition might be problematic, |
| 8 |
> but let's ignore that for now). |
| 9 |
|
| 10 |
I agree more options are needed, but i think a minimal set of defaults |
| 11 |
that user can't change is a good thing (there is no point to change |
| 12 |
most of options that are currently default in fact). Also, options |
| 13 |
like verbosity and progress meter make more sense when controlled on |
| 14 |
command line using -v/-q and --nospinner from emerge than in a |
| 15 |
configuration file. So imho, RSYNC_OPTIONS should more be called |
| 16 |
RSYNC_EXTRA_OPTIONS, with default well documented so that we know what |
| 17 |
are options really left to customization. |
| 18 |
|
| 19 |
Another approach would be to have an /etc/portage/rsync.conf with |
| 20 |
detailed option like the ones currently in make.conf, but more numerous. |
| 21 |
This would ensure that only non-default options are modified by user, |
| 22 |
and would solve the transition problem (because new config files are |
| 23 |
always merged whereas make.conf updates are often ignored). Or does this |
| 24 |
sound too "make.conf.d-ish"? |
| 25 |
|
| 26 |
-- |
| 27 |
TGL. |
| 28 |
|
| 29 |
-- |
| 30 |
gentoo-dev@g.o mailing list |
Archives