Re: tlug: Xatm -- comments?

["Stephen J. Turnbull" <turnbull@example.com>]

>>>>> "Chris" == Chris Sekiya <chris@example.com> writes:

    Chris> On Mon, 5 Apr 1999, Stephen J. Turnbull wrote:
    >> Yes!  Get as much of that crap out of the server itself as
    >> possible.

    Chris> Agreed in principle.  gs ain't exactly a speed daemon,
    Chris> however.

(0) That doesn't matter much if some application wakes up in C locale
    and you can do useful work there while waiting for the Japanese to
    show up elsewhere.  This is not possible with a single-threaded
    X-tt.  There is nothing that stinks more than a dead mouse.  (Boy
    was it a pain in the butt to purge that sucker off my Debian
    system, too.  Everything depends on the X server and vice versa,
    and Debian-JP boofed the pooch on the dependencies....)
(1) If it were done via IPC (which it appears to be) gs's rasterizers
    aren't that slow.
(2) I think the license would allow _you_ to link against the
    Ghostscript graphics library with everything but the rasterizer
    deleted; unfortunately I am rather sure that XFree86 would not be
    allowed to do the same thing (and distribute it).  You could use
    GNU ghostscript instead of Aladdin, of course.  But it would then
    probably have to be a separate server, since it would then have to
    go under GPL.

    >> Ghostscript is not a client here, but functions as a server to
    >> xfs?  That's weird....

    Chris> Just being used as a rasterizer, I think, which makes the
    Chris> diagram somewhat wrong. I'm looking over the code as I
    Chris> write this -- I suspect that each ttf is being broken down
    Chris> into something ps-ish, run through gs, then provided via

It's probably just getting a Type 42[1] wrapper.  This is fast, I
believe.  Internally the engine handles them as TT fonts AFAIK; the
Type 42 wrapper simply makes them palatable to higher-level Postscript
code (in terms of _accessing_ metrics, bitmaps, etc).

    Chris> xfs.  If so, the performance is going to suck.

Not that badly; the code shows that it's piping to and from a
Ghostscript running as a "dedicated server"[2].  See above for the
operations themselves.  Does the gs process get killed after a
reasonable while?

A much bigger problem is the fact that you need to have a Ghostscript
with that particular driver compiled in.  Does the code check, or
could that #ifdef'd stuff go into an infinite loop trying to restart
Ghostscript?  Would it fail with a sensible message?

I assume the missing " in the execlp() arguments is somehow your typo :)

    >> What's wrong with the freetype stuff?  Eg, xfs-tt.

    Chris> Performance problems, political problems.  The usual.

Performance compared to running a separate Ghostscript process?  You
must be joking.

    Chris> Hey, Xatm 

Oh, and the name is a lie; presumably it stands for "X with Adobe Type 
Manager"....


Footnotes: 
[1]  Adobe gallows humor; evidently there was a lot of worry about
TrueType competing with Type 1 fonts, so the Adobe people added a way
to wrap a TT font in Postscript so Postscript code can print using TT
fonts easily.  This, apparently, was the solution to the question of
"Life, the Universe, and Everything."

[2]  Yeh, I know an oxymoron when I write one.

-- 
University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
__________________________________________________________________________
__________________________________________________________________________
What are those two straight lines for?  "Free software rules."
-------------------------------------------------------------------
Next Technical Meeting: April 10 (Sat), 12:30   place: Temple Univ.
*** featuring: LabView and UDB/DB2 for Linux
Next Nomikai: May 21 (Fri), 19:30    Tengu TokyoEkiMae 03-3275-3691
-------------------------------------------------------------------
more info: http://www.tlug.gr.jp        Sponsor: Global Online Japan