Re: check with SGML::Parser::OpenSP (and branches)

On Thu, 2005-08-18 at 15:03 +0900, olivier Thereaux wrote:
> 
> On 18 Aug 2005, at 05:38, Ville Skyttä wrote:
> >
>
> How about adding a paragraph to the installation document, saying:
> [[
>      For better performance, you may enable the template cache  
> mechanism by removing the # character before the file_cache  
> parameter. Note however that this options requires you to run a more  
> recent version of perl (5.8) than the validator usually requires (5.6  
> and above).
> ]]

I had something along those lines in the install doc initially, but left
it out for a few reasons.

It'd be documenting the HTML::Template internals which is better left to
the HTML::Template docs, IMO.  And the above is specific to the
file_cache option; there are other modes that people could want to use
even with the non-mod_perl version, which should probably be documented
as well.  And if we'll require mod_perl later, the default won't be
file_cache, but cache... and file_cache doesn't require Perl 5.8+,
Storable can be installed separately for earlier versions, too.

To fully document this in the validator's docs seems like too much
trouble to me, which is why I tend to think that the commentary and
hints in validator.conf would be enough.

> Also as you noted, 0.8 could quite reasonably require perl 5.8. As  
> far as I can tell, it was available on activeperl since late 2002,  
> and on the mac since 10.3.

Yep, and HEAD already requires it.

> > I'm not sure if the top-level "TemplateOptions" section is the best
> > choice in the config file; if someone has better ideas for the name or
> > the "level", shoot.
> 
> We are in validator.conf, in a context of configuration options. So I  
> suggest simply <Templates>...</Templates>

That could cause problems if we'd like to later add a subsection for
other hypothetical template related options as everything inside the
section will be passed to HTML::Template->new() in the implementation
contained in the patch.  I think this kind of implementation is ok and
future proof, artificially limiting the options that can be specified
there would be a bit silly IMO.

I initially had <Templates><Cache>...</Cache></Templates> and
<Cache><Templates>...</Templates></Cache> but the options in there are
not limited to caching...

<Templates><Options>...?

> > The patch for HEAD would be basically the same as this one, except  
> > that
> > there the default caching would default to "cache = 1", uncommented,
> > assuming that we'll require mod_perl and Perl 5.8.0+ there.
> 
> ok with me. Does anyone know whether we're going to need apache2 +  
> mod_perl2 or if 0.8 will run on 1.3+mod_perl?

I think the question is more whether we want to have a hard mod_perl
requirement in the first place.  mod_perl 1 vs 2 differences (as well as
libapreq 1 vs 2, which would help with getting rid of CGI.pm completely)
are manageable on the fly in the script, despite of the completely
renamed APIs.

Received on Thursday, 18 August 2005 06:38:44 UTC