Re: Framebuffer programming

From: Roger Leigh (${roger}_at_invalid.whinlatter.uklinux.net.invalid)
Date: 10/30/04

  • Next message: Otto Wyss: "Re: Framebuffer programming" 
    Date: Sat, 30 Oct 2004 20:35:53 +0100

    otto.wyss@orpatec.ch (Otto Wyss) writes:

    > Roger Leigh <${roger}@invalid.whinlatter.uklinux.net.invalid> wrote:
    >
    >> otto.wyss@orpatec.ch (Otto Wyss) writes:
    >>
    >> > Ahhh, sorry. The correct URL is "http://linux-fbdev.sourceforge.net/".
    >> > Where did you find the old inactive "www.linux-fbdev.org"?
    >>
    >> In the Framebuffer-HOWTO, while googling for "linux framebuffer
    >> programming". It's in the one linked to from the new site.
    >>
    > Please give me the exact link so I may see that the correct URL is
    > mentioned. Or better send the maintainer a mail yourself, that he should
    > correct it.

    http://en.tldp.org/HOWTO/Framebuffer-HOWTO.html

    (from http://linux-fbdev.sourceforge.net/links.html)

    >> WRT asking on mailing lists, I haven't yet, but I am kind of hoping
    >> that something as important as the framebuffer actually has some
    >> written reference! The new site does not appear to have an API
    >> reference. Are the ioctls and available pixel formats documented
    >> anywhere other than the source? Basically, I just want to understand
    >> <linux/fb.h>.
    >>
    > As much as I know the only description is within the
    > "Linux/Documentation/fb" itself. For more info you have to ask in the
    > mailing list. BTW the old website can be view at
    > "http://linux-fbdev.sourceforge.net/old", if you find some useful
    > material there just say so, so the can be made available again.

    Thanks, but I didn't see anything.

    > It would be nice if a reference would be put onto the web but so far
    > nobody voluntiered to do it. Would you?

    I have written quite a number of manuals, reference and tutorial
    documents, but only when I have been intimately acquainted with the
    subject matter. Since my knowledge of the framebuffer is currently
    minimal, I'm not in a position to help here.

    I don't mean to offend, but the framebuffer has been around for over
    *6* *years*. It's pretty rubbish that *none* of the developers have
    taken the hour or so it would take to document the user-level
    interface in that time... It's doubly rubbish because it's not even a
    big interface. If you want people to actually use the framebuffer,
    then you will have to document it...

    [Example: I used FreeType2 for the first time last week. My first
    program to render and draw glyphs worked perfectly first time. Why?
    Because the FreeType folks have *actually bothered* to write both a
    complete API reference and tutorial. Most folks *do* write API
    references. It needn't be a literary masterpiece, just a description
    of what it does.]

    I /could/ use it by making use of the code examples I've seen (I have
    already got working code), but to write a truly portable program I
    need to fully understand it, and I can see there are endian issues and
    visual issues that I will get wrong (i.e. write broken or nonportable
    code) unless the API and data structures are actually described in
    full. I don't want my program restricted to 32 bpp LE. The docs fb/*
    don't even tell you what the pixel layouts are for each colourmap, let
    alone describe how to set pixel values for each possible variant, and
    I don't like guessing when I'm writing code--I like my code to
    actually work because it was written properly, not because of blind
    luck... 

    -- 
Roger Leigh
                Printing on GNU/Linux?  http://gimp-print.sourceforge.net/
                Looking for a Better Distribution?  http://www.debian.org/
                GPG Public Key: 0x25BFB848.  Please sign and encrypt your mail.
  • Next message: Otto Wyss: "Re: Framebuffer programming"

    Relevant Pages

    • Implementing expected core OS functionality
      ... Most documentation and sample code on Windows ... Dolphin provides a wonderful platform for studying the OS API ... might reference the particular API. ...
      (comp.lang.smalltalk.dolphin)
    • Re: Determine if an item in a collection has changed
      ... What people do with API - I don't know. ... >> parent, which by necessity holds a reference to each of its children (via ... >> hold its pointer. ...
      (microsoft.public.vb.general.discussion)
    • Re: Passing python list from C to python
      ... you are importing it from C, but importing it from Python [so that you ... Python/C API reference is 162 pages! ...
      (comp.lang.python)
    • Re: "Cant return a temporary from lvalue subroutine..."
      ... A similar problem exists with objects that expose their interior though overloading a dereference operator, ... I have used tied hashes to correct this (that is, returned a reference to a tied hash in response to %, and haven't encountered the particular difficulty you're seeing. ... I wrote whole load of in-house classes that used this API technique and then had to labourously go through all our programs to use a direct method-call API rather than the overload-tie API. ...
      (comp.lang.perl.misc)
    • [PATCH 3/5] rcu file: use atomic primitives
      ... * kernel for reference counting. ... * @rcuref: reference counter in the object in question. ... * use the rcuref_inc_lfapi to acquire a reference ... * in a lock-free reader-side critical section. ...
      (Linux-Kernel)