[Pkg-shadow-devel] Re: Debian patches which can be integrated safely

[Christian Perrier]

(better keeping the list CC'ed)

> > 005_manpages.dpatch: trivial addition to useradd.8 for a corner case
> >                      applies cleanly on CVS
> 
> I saw this patch. This patch adds coomment: "This option may not
> function correctly if the username has a / in it.".
> But useradd dissallow create user account with "/" in login name. So
> what is the sense adding this kind of comments ? :)

I'll check more carefully.

> 
> > 204_pt_BR-manpages.dpatch: trivial fixes to pt_BR manpages
> 
> revision 1.4
> date: 2005/11/02 16:03:40;  author: kloczek;  state: Exp;  lines: +5 -2
> merge 204_pt_BR-manpages Debian patch.

OK. I saw this in the meantime.

> 
> > 206_ko-manpages.dpatch: trivial fixes to ko man pages
> 
> revision 1.3
> date: 2004/01/14 14:11:26;  author: kloczek;  state: Exp;  lines: +2 -0
> New version 4.0.4.1
> bug fixes in automake files for generate correct tar ball on "make
> dist":
> added missing "EXTRA_DIST = $(man_MANS)" in man/*/Makefile.am.

Hmmm, that's not what the 206 patch does: it removes newgrp.1 from the
generated man pages. Is that the same than yours?


> 
> > 333_shadow.5-typo_312430: typo in shadow.5
> 
> Something is wrong with this patch:
> 
>        The password field must be filled. The encrypted password
> consists of
> -      13 to 24 characters from the 64 characters alphabet a thru z, A
> thru
> +      of 13 to 24 characters from the 64 character alphabet a thru z, A
>        Z, 0 thru 9, \. and /. Optionally it can start with a "$"
> character.
> 
> Adds double "of" and remove "thru" (?).

Yep, I've corrected it.

> 
> > 401_cppw_src.dpatch: new program cppw (see header)
> 
> Hmm .. if it will be possible I want ommit situation with mixing
> license. Can I add this under the same license as shhadow ?

Well, you need an agreement by Stephen Frost here.

> 
> > 405_subsystem_remove_*_in_shell.dpatch: "does not harm" said Nicolas
> 
> IMO not neccessary now. Any test case ?

I guess no. I hesitated removing it anyway. So removing.

> 
> > 407_32char_grnames_240456.dpatch: allow group names with 32 characters
> 
> It must performed in diffrent way. Instead hardcode group string len
> better will be use sizeof() of some structure.
> The same can be chaged in case urer name (I'm not shure but IIRC in this
> case can be used (sizeof(ut.ut_user)-1) or something simillar).
> So it is for finish (it will be good review all code for hardcoded this
> trings length) byt using RightWay(tm).

OK. We'll see this later.

> > 432_login_cancel_timout_after_authentication: seems fair..:-)
> 
> Can't apply cleanly.

OK, I'll check.


> 
> > 434_login_stop_checking_args_after--: Nicolas said "It could certainly
> >                                       be submitted to upstream."
> 
> Better will be rewrite login for use getopt_long().

OK, comment added.

> > 438_su_GNU_origin: better attribution of some parts of the code
> >                    THIS IS A MUST
> 
> It is neccessary if all other su patches will be appled. Current su have
> modifications for use getopt_long() so all this patches must be
> rediffed.

Sure, but here we talk about code attribution, not really code. So I
don't see what prevents adding a correct atrtibution to authors of
parts of the code.

> 
> > 442_useradd-O_option: better document -O in useradd
> 
> It is Debian specyfic patch for keep some backward compatibilities when
> -O option was not documented (now it is -K and it is documented).
> If there -O isn't used now in wny Debian scripts IMO this patch cen be
> dropped.

There's no more -O option at all in useradd?