Re: [Mason] Bug report (fwd)

[Dave R. <au...@ur...>]

On Thu, 13 Dec 2001, Bojan Jovanovic wrote:

> I use the different parsers for html and for non-html output. However the
> component caching (actually "compiled" components in obj directory) do not
> know what parser the component was built with. The way component was
> created the first time is "recorded":

Let's call those compiled components so as not to confuse them with
Mason's data caching system.

> Maybe instead of having 'h' there you can have a callback to the current
> parser and ask it for the current default encoder...
> Or maybe even better, the components should be "cached" based on the type
> of the parser used (including params sent to the parser)?

This should be largely fixed in 1.10.  In that version, each compiled
component contains a 'compiler_id' that represents what options were used
to compile that component.  If we try to use that compiled component but
that id has changed (because we're using a new compiler) then the
component is re-compiled.

This is still less than ideal because what we should probably be doing is
storing _all_ the different versions.

Also, the compiler id can't deal with things like the pre and post process
callbacks very well (basically, it either knows that they are being used
or not).

But its still a lot better than nothing and should definitely fix the
problem you're seeing.


-dave

/*==================
www.urth.org
We await the New Sun
==================*/