David Kastrup | 1 Sep 2003 18:09
Picon
Picon

Re: [preview-latex-devel] CVS-1.188; Sub-pixel rendering of previews

Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:

> With the current CVS versions of Emacs, AUCTeX and preview-latex the
> previews look like if a sub-pixel rendering was applied to an already
> anti-aliased font/image.  A screenshot showing this with circ.tex can
> be found at http://www.iwi.uni-sb.de/angeli/misc/preview.png.

[Screenshot tells that the sub-pixel rendering has been fixed already
but hangs persist]

I am afraid that this can't really be a preview-latex problem.  The
default preview-latex operation produces PNG files, and the order in
which they are produced from EPS depends on what is on the screen.
If you wait until the "[Compiling]" tag in the mode line goes away,
and then start scrolling around, does Emacs still freeze?  If it
does, then the problem is with the PNG loading code of Emacs.  If it
does this only when the [Compiling] tag is still there, then this
might be a process interaction problem with GhostScript (but then the
freezes should likely also occur when you are scrolling a different
window, one not under the control of preview-latex).

--

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
Ralf Angeli | 1 Sep 2003 18:40
Picon

Re: CVS-1.188; Sub-pixel rendering of previews

* David Kastrup (2003-09-01,18:09) writes:

> If you wait until the "[Compiling]" tag in the mode line goes away,
> and then start scrolling around, does Emacs still freeze?

Yes.  I can reproduce this with circ.tex by doing a `C-c C-p C-d y',
waiting for the compilation to finish and hitting the down arrow on
the keyboard until I reach the line of the file which says "$y$-Wert
dar.  Jetzt formen wir ...".  In this state the rendered equation
below the line is only partly visible.  With the cursor on the
rendered $y$ hitting the down key again produces a hang.  One can
still move the scroll bar but there will be no reaction in the buffer.
The same is true for hitting any of the cursor keys.

> If it
> does, then the problem is with the PNG loading code of Emacs.

I can write a bug report to the emacs-pretest-bug list.  Is there a
possibility to provide a testcase for the list without using
preview-latex?

Cheers,
Ralf

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
Oliver Jennrich | 5 Sep 2003 09:51
Favicon

rel-0-7-8; Preview 0.78 does not work under XEnacs 21.4.12


I'm feeling incredibly stupid as *exatly* this file used to run with
no problems under my old 'preview' installation. Am I missing
something obvious here?

Running preview on the following file:

\documentclass{article}
\usepackage{preview}

\begin{document}

$a+b=c$
\end{document}

yields

Running `Preview-LaTeX' on `test' with ``latex "\nonstopmode\PassOptionsToPackage{active,dvips,auctex}{preview}\AtBeginDocument{\ifx\ifPreview\undefined\fi}\input{test.tex}"''
This is TeX, Version 3.141592 (MiKTeX 2.3)
LaTeX2e <2001/06/01>
Babel <v3.7h> and hyphenation patterns for english, french, german, ngerman, du
mylang, nohyphenation, loaded.

(test.tex (C:\texmf\tex\latex\base\article.cls
Document Class: article 2001/04/21 v1.4e Standard LaTeX document class
(C:\texmf\tex\latex\base\size10.clo)) (C:\texmf\tex\latex\preview\preview.sty
(C:\texmf\tex\latex\preview\prauctex.def
(C:\texmf\tex\latex\preview\prauctex.cfg))
No auxiliary output files.

(Continue reading)

David Kastrup | 5 Sep 2003 10:25
Picon
Picon

Re: [preview-latex-devel] rel-0-7-8; Preview 0.78 does not work under XEnacs 21.4.12

Oliver Jennrich <oliver.jennrich <at> esa.int> writes:

> I'm feeling incredibly stupid as *exatly* this file used to run with
> no problems under my old 'preview' installation. Am I missing
> something obvious here?
> 
> Running preview on the following file:
> 
> \documentclass{article}
> \usepackage{preview}
> 
> \begin{document}
> 
> $a+b=c$
> \end{document}

[Settings]

>  preview-default-preamble ""

Is there a particular reason you set preview-default-preamble to an
empty string?  The default is

'("\\RequirePackage[" ("," . preview-default-option-list) "]{preview}")

If you goofed this up using customize, I recommend that you do
M-x customize-variable RET preview-default-preamble RET
and reset to default settings.

If you set this somewhere in your .emacs, I recommend that you clean
(Continue reading)

Oliver Jennrich | 5 Sep 2003 14:51
Favicon

Re: [preview-latex-devel] rel-0-7-8; Preview 0.78 does not work under XEnacs 21.4.12

* David Kastrup writes:

> If you goofed this up using customize, 

Very probably. OTOH, the 'new' setting looks quite different from my
old one (which didn't work either with 0.7.8, hence the empty entty)

But be it as it might:

> I recommend that you do M-x customize-variable RET
> preview-default-preamble RET and reset to default settings.

Works like a charm.

Thanks
Oliver

--

-- 
Oliver Jennrich                +31-71-565-6074
LISA Project Scientist
ESA/ESTEC, Keplerlaan 1, Postbus 299, 2200 AG Noordwijk, The Netherlands

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
Joachim Schlosser | 26 Sep 2003 21:29

rel-0-7-8; footnote relocated by format (M-q and others)

Hi David and others,

When having activated footnote-previews -- which is the default -- and
then reformatting a block, e.g. with C-c C-q C-p or M-q, the footnote
changes its position in the line and several spaces are inserted. This
of course leads to unwanted effects, namely that a footnote is not
positioned directly after a word but separated with a space in front,
which does not look very nice.

Testcase: Every text with a footnote spanning more that one line.

To produce bug with attached example: 
1. C-c C-c g
2. Move cursor to the paragraph, but without expanding footnote
3. M-q

-----
\documentclass{scrartcl} 
\begin{document}

CTAN.org\footnote{CTAN ist die Abkürzung für \emph{Comprehensive TeX
    Archive Network}, der vollständigsten Sammlung von
  \TeX{}-relevanter Software.}  heruntergeladen werden kann. Unbedingt
notwendig ist auch die De-TeX-FAQ, zu beziehen vom

\end{document}
----

Sincerely,
Joachim
(Continue reading)

Brian V. Smith | 29 Sep 2003 20:37
Favicon

dvipng bugs

Hello,

I just downloaded and compiled dvipng 0.6 and found several problems.
First, it complains:

dvipng warning: Unimplemented color specification ' Black'

Notice there is a blank (space) before the "B" in black.  You should probably 
skip any whitespace instead of assuming there is exactly one blank following 
the commands.

I then edited the .dvi file to remove one of the two blanks and got a 
segfault.
Here is the last output from "ltrace":

ftell(0x08062a80, 0x0805a58c, 100, 0x4207ac7e, 0x08062ea8) = 104
fgetc(0x08062a80)                                 = '\357'
fgetc(0x08062a80)                                 = '\021'
fgetc(0x08062a80)                                 = 'c'
fgetc(0x08062a80)                                 = 'o'
fgetc(0x08062a80)                                 = 'l'
fgetc(0x08062a80)                                 = 'o'
fgetc(0x08062a80)                                 = 'r'
fgetc(0x08062a80)                                 = ' '
fgetc(0x08062a80)                                 = 'p'
fgetc(0x08062a80)                                 = 'u'
fgetc(0x08062a80)                                 = 's'
fgetc(0x08062a80)                                 = 'h'
fgetc(0x08062a80)                                 = ' '
fgetc(0x08062a80)                                 = 'B'
(Continue reading)

David Kastrup | 29 Sep 2003 22:03
Picon
Picon

Re: [preview-latex-devel] dvipng bugs

"Brian V. Smith" <BVSmith <at> lbl.gov> writes:

> I just downloaded and compiled dvipng 0.6 and found several
> problems.  First, it complains:
> 
> dvipng warning: Unimplemented color specification ' Black'
> 
> Notice there is a blank (space) before the "B" in black.  You should
> probably skip any whitespace instead of assuming there is exactly
> one blank following the commands.
> 
> I then edited the .dvi file to remove one of the two blanks and got a 
> segfault.

You can't do that as easily.  Specials in a dvi file contain a length
byte, they are not null-terminated or something.

Of course, for the rest we better wait what Jan-Åke will tell us.

--

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
Jan-Åke Larsson | 30 Sep 2003 09:51
Picon
Picon
Favicon

Re: [preview-latex-devel] dvipng bugs

David Kastrup wrote:
> "Brian V. Smith" <BVSmith <at> lbl.gov> writes:
> 
> > I just downloaded and compiled dvipng 0.6 and found several
> > problems.  First, it complains:
> > 
> > dvipng warning: Unimplemented color specification ' Black'
> > 
> > Notice there is a blank (space) before the "B" in black.  You should
> > probably skip any whitespace instead of assuming there is exactly
> > one blank following the commands.

This is a bug in dvipng 0.6. In the CVS version, I fixed it a month
ago. (some day I'll fix the psnames also)

> > I then edited the .dvi file to remove one of the two blanks and got a 
> > segfault.
> 
> You can't do that as easily.  Specials in a dvi file contain a length
> byte, they are not null-terminated or something.

Removing a blank made the first special (length 17, 
'color push  Black') eat the op-code of the next DVI command (DOWN3).
dvipng then would have interpreted the first byte of the length as an
op-code. I suspect all hell broke loose.

Use the -d switch if you want to know exactly what went wrong. 

Possibly you overran the buffer trying to end the page. I'll implement
sanity checks when I get the time, but just now, Postscript fonts is
(Continue reading)

Jan-Åke Larsson | 30 Sep 2003 23:17
Picon
Picon
Favicon

Re: [preview-latex-devel] dvipng bugs

Jan-Åke Larsson wrote:
> Possibly you overran the buffer trying to end the page. I'll implement
> sanity checks when I get the time, but just now, Postscript fonts is
> the focus.

Correction: sanity checks for the DVI is already there ("DVI ends
prematurely"). What happened to you was that dvipng tried to typeset
the character \001 _before_ any font was selected. I'll add a warning,
but this should never happen unless the DVI is faulty.

/JÃ

--

-- 
C Code. C code run. Run, code, run! (please?)

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Gmane