[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [RFS]: jsMath:TeX equations in HTML documents



Hi Frank,

Thank you very much for your speedy reply and the comments. I want to
discuss some of them though...

> Some comments:

> debian/copyright: I think that a binary deb package is a derivative of
> both the upstream source and the Debian packaging.  According to
> http://www.gnu.org/licenses/license-list.html, upstream's Apache
> license (v.2.0) and your GPL for the packaging are incompatible.
 Tricky point... Indeed GPL and Apache licenses are incompatible
since Apache has some additional conditions (according to the excerpt on
the handy link you've provided).  I don't think that it is required to
have debian packaging material (which is simply the content of debian/
directory) under compatible license with the main content. Simple
arguments by examples:

1. What would be the license of the debian packaging of any non-free
package? :-) they all provide the material which is under GPL
incompatible license, but packaging is DFSG compliant.

2. To don't go into problematic non-free land, quick look found me some
more packages which have their packaging under GPL whenever the rest of
the package (or parts of it ;-)) under Apache (for instance apache-ssl).
Even worse - those packages might combine parts which come from both
Apache and GPL licensed modules.

So, most probably, debian packaging is not considered a derivative from
the original upstream material (especially if it doesn't modify any
upstream files) -- is is barely a container. If my debian packaging
introduced some patches to the upstream, then I had to release those
under Apache license, whenever the rest of debian packaging under GPL...
or am I wrong?

Should we ask on d-legal to clarify the position?

> maintainer scripts:

> - it's inconsistent to name them jsmath.config but postinst and postrm
>   without the package name prepended
agree... all become jsmath.whatever

> - What's the purpose of this in config:
> db_version 2.0 || [ 0 -lt 30 ]
you caught me: I don't know from top of the head. Evil cut&paste. Indeed
2nd clause is effectively call to 'true'.

>   This looks just like || true, and debconf-devel(7) suggests you never
>   need this command...
Hm... Could not see that -- actually it suggest us to use it:

       Notice  the  uses of "|| true" to prevent the script from dying if deb-
       conf decides it can't display a question, or the user tries to back up.
       In  those  situations,  debconf returns a non-zero exit code, and since
       this shell script is set -e, an  untrapped  exit  code  would  make  it
       abort.

> - the config script should try to detect which servers are actually
>   installed, to make sure the user does not select exactly the one
>   that's *not* installed.  Might be tricky to implement, though.
yeah -- tricky... what should I do if the package was removed? so there
is configuration left but no binaries? should I include that one?
should I include preconfigured ones? (probably I should)
Would this selection work, you think?

dpkg -l apache apache-ssl apache-perl apache2 | tail -n +6 | awk '/^[^p]/{print $2;}'

Indeed it would be nice to have it but I see it more of a wishlist
item... ;-)

> - the second debconf question is asked with higher priority than the
>   first (why?), but the text is written with the assumption that the
>   user has already seen the first question.  This should be changed...
higher priority: we can allow default list of web servers so postinst
just corrects all present configuration files, but we really would like
to ask a user to ask whether to restart his webservers.

I've corrected message -- removed "selected"

> - the postinst is probably RC buggy, see policy 10.7.4: "The maintainer
>   scripts must not alter a conffile of any package, including the one
>   the scripts belong to."  At least /etc/apache2/httpd.conf doesn't seem
>   to be a conffile, but I think this mostly applies to configuration
>   files, too.
just only to those that are conffiles. There are some configuration
files which are created by postinst scipts and are totally valid
according 10.7.3.

And it seems you are right -- apache2/httpd.conf is not a conffile since

root@www:/# ls -l /etc/apache2/httpd.conf 
-rw-r--r--  1 root root 268 2006-03-07 01:44 /etc/apache2/httpd.conf
root@www:/# dpkg -S /etc/apache2/httpd.conf
dpkg: /etc/apache2/httpd.conf not found.
root@www:/# dpkg -l apache2
ii  apache2        2.0.54-5sarge1 next generation, scalable, extendable

Hm... so are we RC buggy?

>   Moreover, I doubt that this works at all, and I think there are better
hm... seems to be working to me ;-) I've tested it on a box and
previousely I had gallery installed on a few, that is why I recalled
that it copes with registering itself within apache

>   mechanisms to interact with webserver packages than that.  You write
any pointer would be great -- may be some exemplar package

>   you've taken the code from gallery, but that doesn't mean that it is
>   good.  
it might be not the best but it is working one ;-)

> debian/rules:

> - why do you call dh_strip and dh_link?
Indeed... leftovers -- thanks for spotting! removing them from all
jsmath* packages

> So much for this package, Frank
Thank you for the comments.

P.S. Since modifications were quite minor I didn't boost up a revision,
so the files are available from the same URLs

-- 
                                  .-.
=------------------------------   /v\  ----------------------------=
Keep in touch                    // \\     (yoh@|www.)onerussian.com
Yaroslav Halchenko              /(   )\               ICQ#: 60653192
                   Linux User    ^^-^^    [175555]


Attachment: pgpYCeTFRfjGw.pgp
Description: PGP signature


Reply to: