On Thu, May 31, 2018 at 10:32:51AM +0200, Tomas Mraz wrote:
On Thu, 2018-05-31 at 06:14 +0300, Dmitry V. Levin wrote:
> Some of prompts issued directly or indirectly by pam_unix that start
> with a capital letter are finished with a dot, some are not.
> Fix this grammatical inconsistency by changing all these prompts
> to finish with a dot.
>
> * modules/pam_cracklib/pam_cracklib.c (_pam_unix_approve_pass):
> Harmonize prompts.
> * modules/pam_unix/pam_unix_acct.c (pam_sm_acct_mgmt): Likewise.
> * modules/pam_unix/pam_unix_passwd.c (_pam_unix_approve_pass,
> pam_sm_chauthtok): Likewise.
> * po/Linux-PAM.pot: Regenerate.
Is this really worth the change of translatable strings triggering
needed changes in all the translations? I'd prefer at least postponing
this commit until we have other more significant changes in translated
strings such as new translatable strings added or so.
There is a background for this change that is not quite obvious so I'd
rather tell the full story.
I maintain an out of tree module called pam_tcb, it's a part of tcb suite.
This module was born as a fork of pam_unix to implement the tcb password
shadowing scheme about 17 years ago, i18n support was added there 7 years
ago, and less than a year ago I've got a bug report that some messages are
untranslated because pam_tcb uses "Linux-PAM" domain and some prompts have
this "trailing dot" difference with pam_unix.
Now I have to choose whether to fix Linux-PAM translations or to fork them
into pam_tcb and maintain them there. As the second approach would result
to worse quality of translations, I'd rather follow the first approach.
For this I need a consensus that the change of translatable strings is OK
and will be applied before the next release of Linux-PAM whenever it
happens.
--
ldv