[schooltool-dev] Re: [SPAM] Re: [Cando] Re: Upcoming sprint adjustments

[Tom Hoffman]

On 6/22/07, Paul Carduner <paulcarduner at gmail.com> wrote:
>
> After more work with SIF stuff, I am again feeling somewhat
> discouraged.  It seems like both the sifagent that Will wrote and the
> agent framework (Chesty) that Tom wrote are in serious need of
> refactoring.  For example, in the file
>
> http://codebrowse.launchpad.net/~chesty-devs/chesty/chesty-trunk/annotate=
/paul%40carduner.net-20070619041842-ke2juio8hmlx0axu?file_id=3Dsif_exceptio=
ns.py-20070329154951-he3tabno2t3clr7y-129
> we "could reduce the amount of code in there like 20 fold." (ignas)


In defense of my code, the exceptions are a particular case.  I'm sure you
could condense that down quite a bit, but it really wouldn't make any
difference.  It would be just a more consise way of creating a bunch of
class-based exceptions required by the spec.

In order to really understand what this code is doing, we will have to
> refactor it a lot.  Before embarking on the refactoring journey, I
> tried running the chesty tests and just got back:
> WARNING:root:SIF Error cat. 1, code 2: Message is not well formed.
> WARNING:root:SIF Error cat. 3, code 5: Sender's certificate is not
> trusted.
> which is also discouraging.  The doctests also don't do a great deal
> of explaining about what is going on.


While the doctests don't do a great deal of explaining what is going on,
they do note that they are sending messages that should be creating these
errors, which are logged to the console.  Perhaps they shouldn't be, but
that's turned off easily enough in the configuration file.


> Anyhow, if we end up doing SIF stuff during the sprint, we will
> probably spend the whole time refactoring the existing code - which
> isn't necessarily a bad thing as it will have to be done at some point
> or another.  But again, i'll just reiterate that the presence of Tom
> and WIll would *really* help in a refactoring effort.


The fact of the matter is that the code basically works, and if there is
something fundamentally wrong with it, I'd like to know what the problems
are.  I think we could move forward if:

a) figure out how to make changes made by stAgent properly trigger
SchoolTool events;
b) figure out how to run chesty directly from the SchoolTool process.

Does anyone else have other thoughts on this matter?  I still haven't
> started thinking about the REST/site.zcml/threading issue, as the code
> is not in the right state for "messing around".
>
> Does anyone have any other thoughts on the matter?


I think I don't even know why I would try to contribute to the project I
manage when I get this kind of response from the developers, let alone
anyone else contributing.

Regardless, this is probably not going to work.

--Tom
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.schooltool.org/pipermail/schooltool-dev/attachments/20070=
623/f20d3ce4/attachment.htm