[tg-trunk] Future of Genshi

[Alessandro Molina]

I'm writing here because I'm fearing we will have to make a clear step
regarding genshi in the near future. So I would like to gather as much
opinions as possible regarding the matter.

Currently we have a major issue with TurboGears on Python3.4 caused by
Genshi not working there due to changes into the ast module. The issue has
actually been already patched nearly and year ago in genshi itself, but no
new releases ever happened.

I have been able to provide a genshi.name_constant_patch option in TG
genshi renderer that monkeypatches it to make it work on Py3.4, but this is
actually a temporary work-around and is far from being an acceptable
solution.

I also got in touch with Simon Cross half an year ago and he confirmed he
was willing to make a new genshi release, but it still didn't happen.

Any suggestion on how to proceed?
I'm currently only able to think of replacing Genshi with Kajiki as the
official TG template engine in a future major release. I'm unable to
maintain genshi itself as its codebase is quite complex and hard to
maintain, while Kajiki is 90% compatible with genshi while having a much
simpler codebase and faster performances.

This is off course a last resort and I would really like to avoid having to
proceed that way, but if Genshi is not maintained anymore and it stops
working on future python releases I'm currently unable to see a better move
:/

[Christoph Zwerschke]

Post by Alessandro Molina
I'm writing here because I'm fearing we will have to make a clear step
regarding genshi in the near future. So I would like to gather as much
opinions as possible regarding the matter.

First of all, it's very sad that Genshi has stalled. The idea behind it
(actually copied from "Kid") was brilliant, particularly at that time
when XHTML was all the rage. Now with HTML5 the use of XML has become a
bit less interesting. But Genshi worked still quite well for me as a
successor of Kid.

Unfortunately, I can't give feedback regarding Kajiki, since I haven't
used it so far. It doesn't seem very active either, and doesn't seem to
have many users, so there may be lingering problems that nobody noticed
so far. The CDATA bug you mentioned is still not fixed, either. TW2
compatibility is very important.

It seems your argument is that Kajiki can be easier forked and/or
maintained as part of the TurboGears project, because it's less complex,
do I understand it correctly?

-- Christoph

[Alessandro Molina]

Post by Christoph Zwerschke

Post by Alessandro Molina
I'm writing here because I'm fearing we will have to make a clear step
regarding genshi in the near future. So I would like to gather as much
opinions as possible regarding the matter.

First of all, it's very sad that Genshi has stalled. The idea behind it
(actually copied from "Kid") was brilliant, particularly at that time when
XHTML was all the rage. Now with HTML5 the use of XML has become a bit less
interesting. But Genshi worked still quite well for me as a successor of
Kid.

Yes, and I really believe it is one of the foundation concepts behind TG.
the XML based template engine is one of the features I love most and
that differentiates TG from the other frameworks, so we definitely had to
retain it.

Post by Christoph Zwerschke
Unfortunately, I can't give feedback regarding Kajiki, since I haven't
used it so far. It doesn't seem very active either, and doesn't seem to
have many users, so there may be lingering problems that nobody noticed so
far. The CDATA bug you mentioned is still not fixed, either. TW2
compatibility is very important.
It seems your argument is that Kajiki can be easier forked and/or
maintained as part of the TurboGears project, because it's less complex, do
I understand it correctly?

Yes,
I actually don't want to replace Genshi at all, I like it but I'm afraid we
won't get the new release any time soon.
As I'm unable to take over the effort of maintaining genshi and the
codebase looks quite complex the only alternative I have been able to think
of is switching to Kajiki which seems to have an easier codebase ( I have
been able to provide pull requests for it in the past ).

Any other suggestion is appreciated :)

[Craig Small]

Post by Alessandro Molina
I'm writing here because I'm fearing we will have to make a clear step
regarding genshi in the near future. So I would like to gather as much
opinions as possible regarding the matter.

I use mako myself, or am I missing the point here?
Unless there is good reason not too, switch to the easiest one to
maintain.

- Craig
--
Craig Small (@smallsees) http://enc.com.au/ csmall at : enc.com.au
Debian GNU/Linux http://www.debian.org/ csmall at : debian.org
GPG fingerprint: 5D2F B320 B825 D939 04D2 0519 3938 F96B DF50 FEA5

[Alessandro Molina]

Post by Craig Small

Post by Alessandro Molina
I'm writing here because I'm fearing we will have to make a clear step
regarding genshi in the near future. So I would like to gather as much
opinions as possible regarding the matter.

I use mako myself, or am I missing the point here?

The core issue is that Genshi has been the standard of TurboGears for the
past 5 years or so,
most applications have been based on Kid and then ported to Genshi, so it
would probably be an issue for many users.

But the most problematic part is it was the template pluggable apps and
extensions could rely on,
independently from the template you quickstarted with,
master.html was provided and genshi available so any pluggable or extension
you used would work even though your app was with mako or jinja.

[Mengu]

as the 20th century poet ozzy osbourne once written: "let it go, let it
die."

Post by Alessandro Molina
I'm writing here because I'm fearing we will have to make a clear step
regarding genshi in the near future. So I would like to gather as much
opinions as possible regarding the matter.
Currently we have a major issue with TurboGears on Python3.4 caused by
Genshi not working there due to changes into the ast module. The issue has
actually been already patched nearly and year ago in genshi itself, but no
new releases ever happened.
I have been able to provide a genshi.name_constant_patch option in TG
genshi renderer that monkeypatches it to make it work on Py3.4, but this is
actually a temporary work-around and is far from being an acceptable
solution.
I also got in touch with Simon Cross half an year ago and he confirmed he
was willing to make a new genshi release, but it still didn't happen.
Any suggestion on how to proceed?
I'm currently only able to think of replacing Genshi with Kajiki as the
official TG template engine in a future major release. I'm unable to
maintain genshi itself as its codebase is quite complex and hard to
maintain, while Kajiki is 90% compatible with genshi while having a much
simpler codebase and faster performances.
This is off course a last resort and I would really like to avoid having
to proceed that way, but if Genshi is not maintained anymore and it stops
working on future python releases I'm currently unable to see a better move
:/