I got a discussion with Nicolas today about the slow maintenance of the
Genshi project. The last release was made 2 years ago and we have been
affected by  which is fixed but not in any published release.
So the first idea was to maybe migrate to Jinja2 but there are many
issues with such migration. Jinja2 is not XML based so it is much more
complex to manipulate the ODT XML. Jinja2 doesn't allow to call methods
(almost) but Relatorio adds many method call in the instruction for
cache, counting etc.
So the next idea was to see why the Genshi project was not active and
see if we could help. I checked the mailing list and indeed we are not
alone  and Genshi seems to want to migrate to GitHub before doing any
release but this seems to progress very slowly.
In this discussion, I found Kajiki  which is an XML base template
engine using almost the same syntax as Genshi. It seems to be much more
actively maintained. They also have a Text base template engine which is
nice for the non OpenDocument stuffs. They also have a page about
migrating from Genshi  so they clearly position themselves as an
The drawbacks I see for now, is the PyPI stats showing 10 times less
download as Genshi and there are no filters (except the i18n feature).
(Tryton uses only i18n filters)
I think it worths trying to do such switch.
If you agree I can propose this idea to the list of Tryton ideas for
the GSoC .
Cédric Krier - B2CK SPRL
Tel: +32 472 54 46 59