Understanding how one issue relates to others would be easier if issue links had
the linked issue's title as a tooltip. So we'd have:
    <a href="issue6020" title="Create a datetime.timedelta.totalseconds
property">#6020</a>

We can easily add this to the dependencies and superseder links. Adding titles
to issue links in message bodies is a bit more involved, but doable. As a
benefit, we'd be able to avoid linking to bogus issues.

RFE by Ezio Melotti and R. David Murray.

The attached patch:
 * Fixes an 'integer too large' error in templating.py that causes issues to be
undisplayable.
 * Adds title attributes to Roundup's class links (e.g. 'version 1').
 * Moves in-message issue links of the form 'issue 123' from templating.py to
local_replace.py.
 * Adds issue titles to issue links in dependencies and message bodies.
 * Adds the 'open' or 'closed' CSS classes to issue links. 

I can provide the bugfix part as an independent patch.

Demos:
 * Features: http://bot.bio.br/python-dev-exp/issue14
 * Bugfix: http://bot.bio.br/python-dev/issue11

Just the bugfix.

As suggested by Stefan Seefeld, this patch handles the overflow on the DB layer.

Daniel Diniz wrote:
> Daniel Diniz <ajaksu@gmail.com> added the comment:
>
> As suggested by Stefan Seefeld, this patch handles the overflow on the DB layer.
>   
> --- a/roundup-src/roundup/backends/rdbms_common.py	Tue May 26 15:50:15 2009 -0300
> +++ b/roundup-src/roundup/backends/rdbms_common.py	Wed Jun 03 10:24:15 2009 -0300
> @@ -1110,6 +1110,8 @@
>              # Return 1, not True, to match the type of the result of
>              # the SQL operation below.
>              return 1
> +        if int(nodeid) > 2**30: # Avoid Postgres error
> +            return 0

Yes, I think this is much better. However, I'm still concerned about two 
things:

1) You put postgresql-specific code into rdbms_common.py
2) It seems there may be other cases where we don't properly check for 
type / value incompatibilities between Python and SQL backends.

I'm not sure what to suggest, as I don't know the domain well enough. 
But I wonder: might it be a good idea to have back_postgresql do more 
work, such as provide its own 'sql()' function, where it could either 
validate the arguments, or catch backend-specific errors and translate 
them into Roundup exceptions ?

Thanks,
       Stefan

-- 

      ...ich hab' noch einen Koffer in Berlin...

For reference, traceback is here:  http://paste.pocoo.org/show/120779/

1) rdbms_common.Database.sql is supposed to work with Postgres and MySQL, I'm
just fixing it for the former.
2) Probably yes, but FWIW this patch fixes many other Postgres overflows in
hasnode: for superseder, status, etc.

I think we should fix this, then maybe refactor the DB code. I'm not sure
investing in refactoring and improving our own DB-compatibility layer is
desirable over picking an external ORM.

I also suggest to add in the title="" the status of the issue, something like
title="[open|closed|pending] issue title here".

The alternative is to use a CSS class (as suggested by ajaksu2) and
text-decoration: line-through; or some color for closed issue, but imho this
just introduces useless noise. Even if in this way the user can see immediately
if the issue is closed or not he still has to read its title in order to know
what the issue is about.
(Setting the class="" is still a good idea, even if it's not used for this purpose.)

> The attached patch:
>  * Fixes an 'integer too large' error in templating.py that causes issues to be undisplayable.

How do you get this error exactly?

> * Adds title attributes to Roundup's class links (e.g. 'version 1').
> * Moves in-message issue links of the form 'issue 123' from templating.py to
local_replace.py.

I'd rather see the opposite. IMHO roundup should add a link to #NNNN too. In local_replace there should be only Python-specific replaces (like links to the files and revisions).

> * Adds issue titles to issue links in dependencies and message bodies.
> * Adds the 'open' or 'closed' CSS classes to issue links.

Fixed in r83129, r83130, r83131, and r83132 (see also http://issues.roundup-tracker.org/issue2550659).