On Jul 6, 2011, at 8:54 AM, Hannes Magnusson wrote:

> On Wed, Jul 6, 2011 at 17:15, Philip Olson <phi...@roshambo.org> wrote:
>> 
>> On Jul 6, 2011, at 4:20 AM, Hannes Magnusson wrote:
>> 
>>> On Wed, Jul 6, 2011 at 12:20, Hannes Magnusson
>>> <hannes.magnus...@gmail.com> wrote:
>>>> Hi all
>>>> 
>>>> 
>>>> Currently our php.net/indexes is "manually generated" with the
>>>> genfuncindex.php script..
>>>> 
>>>> I'd like to propose to use an PI for this, and generate the list in PhD...
>>>> You could print out indexes for pretty much anything PhD indexes.. so
>>>> the following patch prints out the list of all refentries and
>>>> examples.
>>> 
>>> 
>>> There is a slight problem with that.
>>> There doesn't seem to be any correct way of differentiating
>>> refentries.. so if its a refentry for function, method, variable, url
>>> wrapper, context option.. PhD has no idea.
>>> I played around a little trying to teach the indexer the difference
>>> (depending on the refnames, and the rest of the structure of the
>>> file), but it becomes very messy and not 100% accurate.
>>> 
>>> I can't think of any other way then to modify the markup to explicitly
>>> mention what kind of refentry it is.. thoughts?
>>> Maybe its not important for the indexes page, just list all refentries
>>> alphabetically?
>> 
>> Wouldn't we use PhD_IDE for this task? It contains the appropriate 
>> information, and has already generated funcsummary.txt. I forgot about 
>> funcindex.xml but will look into doing this.
>> 
> 
> 
> How would that work exactly?
> First run PhD_IDE to generate doc-base/funcindex.xml manually every
> week and then run PhD_PHP like normal?
> 
> And how does PhD_IDE know what is what without scanning the ID and
> blindly guessing?

Sorry, I see PhD_IDE as a magic bullet that solves all problems but shouldn't 
think like this. PhD_PHP should solve the problem but I lack solutions. And 
yes, PhD_IDE generating funcindex.xml was the idea there. Maybe Moacir has a 
few thoughts on this matter.

Also, changing funcindex.xml to a full index is possible. I for one never use 
funcindex, but maybe others like it? A real index may be more valuable but I'm 
not sure how large that'd be, or how it'd look. Ideas? I think it's worth 
pursuing. 

Regards,
Philip

Reply via email to