Replacement for libdbi

Sébastien Villemot sebastien at debian.org
Sat Jan 11 17:10:21 EST 2014


Le samedi 11 janvier 2014 à 13:51 -0800, John Ralls a écrit :

> That said, what I proposed was to fork libdbi or SOCI *into gnucash*. It would become part of the GnuCash source code and would cause you no more trouble than the other bits of GnuCash lifted from other projects like LibQOF or the register (lifted largely from Gnumeric). It might cause *us* some trouble maintaining the code, and your comment is very helpful to that aspect of the discussion. Thanks.

Thanks for your reply. I had actually understood that you meant a fork
inside gnucash. Whether you maintain it within the same source tree or
in a separate source tree is not really relevant in this case.

The problem is rather the duplicate codebase: from the point of view of
the distribution (Debian) as a whole, it really means that there would
be two copies of libdbi in it, and it is bad for the same reasons that
you don't want two almost identical copies of a file in a single project
(you rather want to factorize the common parts, which is the purpose
achieved by a library at the distribution level).
For example, that means that if a security issue were discovered in
libdbi, then our security team would have to analyse and possibly fix
two packages instead of one, hence double workload.

The fact that GnuCash already has embedded copies of libqof and the
gnumeric register is not a good reason to add more embedded copies :)

-- 
 .''`.    Sébastien Villemot
: :' :    Debian Developer
`. `'     http://www.dynare.org/sebastien
  `-      GPG Key: 4096R/381A7594

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://lists.gnucash.org/pipermail/gnucash-devel/attachments/20140111/a438c278/attachment.sig>


More information about the gnucash-devel mailing list