[Fenfire-dev] Answers to humppake

[Tuomas Lukka]

Saw some questions on IRC, but you were gone. Answering here.

        <humppake> Onkohan kuinka vaikeaa muuntaa tuplageometria jakamaan ruutu
        vaakasuunnan  sijaan pystysuunnassa?
        <humppake> Vaihtoehtoisesti tietenkin.  
        <humppake> Ajattelen näitä
        meidän kahden vierekkäisen näytön ratkaisuja, jollaisen olen nyt  saanut
        kotiinkin.
        .

Changing doublegeometer for vertical or any kind of split is trivial.

Also, if you want, you can make two windows, one for each physical screen.
I'm planning to do that for myself at some point.

        <humppake> ... MM:ssä ... lisäksi
        suurin ongelma on  koordinaatistojen luonnin hitaus. 
        ... toinen ratkeaa  kirjoittamalla ainakin
        kaikki mindMapView2D:sta ajatuksen kanssa uudelleensuunnitellen Javaksi.

In view performance, merely translating to Java won't necessarily solve
things. Some vital principles:

        1) *DO NOT* create vobs at view making time without caching.
           Creating vobs is relatively slow as it goes to JNI.
           One to ten vobs per view is ok, but hundreds goes too slow.

           Especially since they will have to be garbage-collected later.

           See e.g. CachedNodeFunction for caching text placeables.

        2) Try not to create objects for keys. That's another strain on the GC
           and wasted time. It would be good to use only objects that are
           already there for the keys.
        
        3) Do not use getSqSize() and transformPoints() when generating the view
        
        4) Post your innermost loop here for professional dissection services ;)

I've done a lot of work to make it possible to make fast code using libvob.
However, it's still **easy** to make slow code by misusing it.

        <humppake> tuomasl2: En onnistunut kääntämään convertilla (ImageMagic)
        kuvaa sellaiseksi  PostScriptiksi, että FenPDF olisi sen hyväksynyt.
        Oletko onnistunut? Millä parametereilla?  Olisiko kuvien lisäämisen
        mahdollisuus "Import PS/PDF":n taakse liikaa tähän versioon? 
        (toisaalta, akuuuttia tarvetta ei ole, jos sen voi convertilla kiertää)

See org/fenfire/fenpdf/BUGS

It currently really accepts only pdf.

As for accepting image files, I don't think we have the appropriate code
in org.fenpdf.spanimages yet - feel free to add. It would be nice to have it,
definitely, and it wouldn't be too complicated to add since it's mostly 
a matter of

        1) getting image type (we have some code for that, I think,
           based on first bytes)
        
        2) getting image size (ditto)

        3) loading the image into a texture when making the compressed textures

That's all.

        Tuomas