Re: [Gzz] Tuukka 2002-10-13 (genimages, jython)

[Tuomas Lukka]

On Mon, Oct 14, 2002 at 01:38:47PM +0300, Tuukka Hastrup wrote:
> On Mon, 14 Oct 2002, Tuomas Lukka wrote:
> > pdfinfo sounds good.
> > We also need the same for .ps, naturally...
> 
> I found no similar tool. Maybe we should write a simple PS DSC info parser 
> and refine it when need arises.
> (http://partners.adobe.com/asn/developer/pdfs/tn/5001.DSC_Spec.pdf)

Sounds reasonable. Preferably python/jython.

> > > Finding libraries to read and render pdf would be even better as we could
> > > get rid of some external program dependencies.
> > 
> > Yes, but I think for 0.8 final acceptable would be
> >     - lazy generation of images
> >     - using pdfinfo and something corresponding for .ps to
> >       get sizes.
> >     - using external programs to generate images
> 
> An intermediate I've been targeting is to allow running the client without
> any external programs present but their output cached. This way, it would
> suffice to have the external programs somewhere, and the cached output
> could be copied to Macs etc.

Argh, that's of course true.

Hmm. 

> This in mind, I think we should include some description files in the
> mstmpdir (We could call it a cache). They would contain the number of
> pages and dimensions of the corresponding mediaserver block. (Maybe also 
> document title to be shown while images are being created.) Client would 
> read these files and the image files, and would try to generate them only 
> if they are not available.

Sounds very good.

> For pdf files, pdfinfo can provide the info. For ps files, we can probably
> trust the DSC comments. For any format, we can find out the info from the
> generated image files - the problem is we don't want to wait for the info
> the time the conversion process takes.

Yes, and also: gs does barf on some of the pdf files. Which means that
there, nothing can be trusted.

        Tuomas