[rt-users] Possible Mason bug (was: Re: [Mason] <%filter> in topmost autohandler bypassed in RT.)

[Emanuele Zeppieri]

Upon further investigations, this unfortunately seems to be a Mason bug
(or at least a documental hiatus) since I switched back to Mason 1.32
(just to test) and everything worked as expected: the filter section in
the topmost autohandler received *all* the generated output (despite the
presence of the various $m->flush_buffer in the other lower level
components).

So, if not a bug, this is definitely a change in the way Mason
autohandler works in conjunction with flush_buffer (a change introduced
somewhere between versions 1.32 and 1.35), which is probably unintended
(or at least undocumented).

Can you please check this?

Thank you,
Emanuele.

> -----Original Message-----
> From: rt-users-bounces at lists.bestpractical.com 
> [mailto:rt-users-bounces at lists.bestpractical.com] On Behalf 
> Of Emanuele Zeppieri
> Sent: Saturday, November 18, 2006 6:32 AM
> To: rt-users at lists.bestpractical.com
> Subject: [rt-users] (Mason related question) <%filter> 
> section in the toplevel autohandler does not work as expected.
> 
> 
> Hi!
> 
> This is probably a Mason question, not an RT question, but since I'm
> completely new to both Mason and RT, and this problem 
> happened to me for
> the first time just tonight, right after my very first RT installation
> and subsequent Mason experiments, you can probably easily 
> reproduce the
> problem or point me to the right direction (that is, let me abuse of
> your competence about Mason, and your kindness).
> 
> I placed a <%filter> block in the topmost RT autohandler component
> (/opt/rt3/share/html/autohandler in my layout) and I expected 
> to be able
> to filter any HTML code generated by RT, instead the
> <%filter> block got almost completely bypassed, that is, upon the
> (unique) filter activation $_ contained only the few last 
> HTML lines of
> the generated page (thus even $_ = '' inside the <%filter> section did
> basically not alter the page).
> 
> After a lot of digging in the Mason docs, I've tried removing some of
> the various $m->flush_buffer scattered over the others RT's Mason
> components, and this seemed to alleviate (if not solve) the 
> problem, in
> the sense that this way I could get much more HTML code in $_ 
> inside the
> filter.
> In other words it seems that $m->flush_buffer bypasses the <%filter>
> block in a higher level component, which is different from what the
> Mason docs state, if I understood them.
> 
> Also consider that my local directory is empty.
> 
> I've also searched through the Mason bug-reports, and this one, in a
> sense, seems to be the opposite of an old Mason bug (now fixed), which
> caused $m->flush_buffer to be a no-op in presence of a
> <%filter> block...
> 
> Could you please check this problem, or tell me if I'm making some
> stupid mistake?
> 
> (I'm just curious about the way Mason works, and I have no 
> intention to
> customize RT this way ;-)
> 
> My setup is:
> 
> RT          3.6.2 RC1 (fresh installation, no customizations)
> HTML::Mason 1.35 
> perl        5.8.8
> Apache      2.2.3
> mod_perl    2.0.2
> Mac OS X    10.4.8 
> 
> BTW, RT is great! :-)
> 
> Thank you,
> Emanuele.
> 
> 
> _______________________________________________
> http://lists.bestpractical.com/cgi-bin/mailman/listinfo/rt-users
> 
> Community help: http://wiki.bestpractical.com
> Commercial support: sales at bestpractical.com
> 
> 
> Discover RT's hidden secrets with RT Essentials from O'Reilly Media. 
> Buy a copy at http://rtbook.bestpractical.com
>