1 |
On Tuesday, July 12, 2011 00:54:02 William Hubbs wrote: |
2 |
> On Mon, Jul 11, 2011 at 11:27:59PM -0400, Mike Frysinger wrote: |
3 |
> > i dont think we've had openrc in stable long enough to force people to |
4 |
> > migrate. so i'd keep the code putting along for now, and add a note to |
5 |
> > the feature removal schedule only after we have checks in the openrc |
6 |
> > ebuild that to catch upgrade paths (and perhaps even do the migration |
7 |
> > for users). |
8 |
> |
9 |
> We don't try at all to migrate /etc/conf.d/net; we tell the user to read |
10 |
> /usr/share/doc/openrc-*/net.example and do the migration themselves. Are |
11 |
> you suggesting automating that step? I think it would be tricky to do |
12 |
> that at best since there could be so many different configurations. |
13 |
|
14 |
it would be nice to come up with some level of automated conversion. no |
15 |
matter how much we say "read the docs", there will always be people who do |
16 |
not. i think the reason the openrc upgrade has gone so smooth thus far is |
17 |
because how much we've done to auto-migrate settings. |
18 |
|
19 |
> Is it possible to use eerror inside the _flatten_array and _get_array |
20 |
> functions some how so that they will complain when they are used to |
21 |
> parse bash arrays? |
22 |
|
23 |
you can add an ewarn now while continuing to parse the code. but the same |
24 |
issue as above ... there are many systems which are headless where these will |
25 |
not generally be seen. |
26 |
-mike |