[arch-dev-public] pending initscripts tasks

[Thomas Bächler]

Roman Kyrylych schrieb:
> OK, I've attached the patch below.
> 
> I've successfully tested it in the following configurations:
> 1)
> LOCALE="uk_UA.utf8"
> KEYMAP="ua-utf"
> CONSOLEFONT=cyr-sun16
> CONSOLEMAP=
> 2)
> LOCALE="uk_UA.koi8-u"
> KEYMAP="ua"
> CONSOLEFONT=koi8u_8x16
> CONSOLEMAP=
> 3)
> LOCALE="ru_RU.koi8-r"
> KEYMAP="ru"
> CONSOLEFONT=Cyr_a8x16
> CONSOLEMAP=koi2alt
> 4)
> LOCALE="uk_UA.utf8"
> KEYMAP="ua-utf"
> CONSOLEFONT=LatArCyrHeb-16
> CONSOLEMAP=
> 
> I would like to hear a confirmation that original report in
> http://bugs.archlinux.org/task/7641 is fixed too.
> 
> Few notices:
> 1) It seems that "\e(K" is non needed for UTF-8 mode, right now it is
> appended to locale.sh in when CONSOLEFONT is set, regardless of
> LOCALE.
> But it doesn't hurt and I didn't want to complicate our if/then/else magic more.
> 2) *If* we'll have localized dmesg, rc.* output or /etc/issue in
> future - we'll need to echo "\e%G" or "\e(K" for current (or, for
> /etc/issue - all consoles) in rc.sysinit too, because right now
> non-Latin chars will get correctly displayed only after login.
> I think we don't need this because a) we don't have localized dmesg
> and rc.* output and b) if user wants non-Latin chars in /etc/issue -
> he can incorporate "\e%G" or "\e(K" codes directly in /etc/issue.
> 3) With some fonts (e.g. LatArCyrHeb-16) everything printed on screen
> before setfont is broken - I couldn't find the way to avoid this. IMO
> this is a non-issue (at least noone even complained about this ever).
> All 3 are not bugs.

I applied your patch. However, doesn't your patch (re)introduce a
problem with non-latin characters in user names? The reason we set the
UTF8 consoles was that login wouldn't use UTF-8 otherwise.

http://projects.archlinux.org/git/?p=initscripts.git;a=summary

>> I remember runlevel switching issues (switching from 3rd to 1st to 3rd
>> level is wrong) but it's not very important (exists for more than year
>> already) because almost nobody does this.
> 
> Why not shutdown *all* daemons in reverse order, not just those that
> are not in DAEMONS?

That breaks if you restart a daemon after you booted, because it will be
shut down in the beginning. This will hurt if you restart network,
because umounting network filesystems will hang indefinitely.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 252 bytes
Desc: OpenPGP digital signature
URL: <http://archlinux.org/pipermail/arch-dev-public/attachments/20071021/dbbc56cf/attachment.pgp>