MetaPost 1.501 (alpha version)
I am happy to announce a new *alpha* release of MetaPost:
----------------------------------------------------------------------
MetaPost 1.501
----------------------------------------------------------------------
The standalone distribution package and a win32 binary can be downloaded
immediately from
http://foundry.supelec.fr/gf/project/metapost/
This is an *alpha* release following a major internal rewrite, so do
not use this version for production purposes, but please test with
as many files as you can.
The win32 binary is intended for texlive or a similar web2c-based
installation, this executable will *not* work for miktex.
Of course, we welcome any comments (either good or bad) that you may
have. Simply replying to this message is fine, but see the bottom of
the message for a more formal way to report bugs and feature requests.
----------------------------------------------------------------------
What is new in MetaPost version 1.501:
----------------------------------------------------------------------
* The maximum number of variable instances has increased from approx.
33 million (2^25-1) to 2 billion (2^31-1).
* New build system and kpathsea imported from TeX Live.
* Many small bugfixes to 1.500.
----------------------------------------------------------------------
What is new in MetaPost version 1.500:
----------------------------------------------------------------------
* Actual memory dump files do not exist any more. --mem=<MEM> and
all other methods of supplying a MEM file are now interpreted
somewhat like this pseudo-metapost-code:
begingroup;
let dump = end;
input MEM[.mp];
endgroup;
this process is supposed to be transparent besides the facto that
there is some extra terminal output and logging.
* All memory is now allocated and freed dynamically. There are only
two remaining overflow errors(1):
"expansion depth" : this is a guard against infinite recursion,
the value is 10000.
"sizes per font" : this is maxed out for data-structure reasons,
the value is 32767.
from texmf.cnf, now only the "error_line", "half_error_line" and
"print_line settings" are used.
(1) but now you can reach the operating system limits, of course.
* As this is an alpha release containing lots and lots of new code,
MetaPost starts up with the internal equivalent of 'loggingall;'
in effect. This results in log files that can help in debugging.
* Expect bugs and (minor) memory leaks.
----------------------------------------------------------------------
Legal notice / license / bug reports
----------------------------------------------------------------------
MetaPost is a system for producing graphics developed by John Hobby
at AT&T, based on Donald Knuth's Metafont. The MetaPost home page
is http://tug.org/metapost.
MetaPost is free software: you can redistribute it and/or modify it
under the terms of the GNU Lesser General Public License as published
by the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
The MetaPost development project is currently hosted at Supelec,
http://foundry.supelec.fr/gf/project/metapost; with a bug tracker
at http://tracker.luatex.org/. Please report bugs and
request enhancements through there if possible. Alternatively, you
can reach us through the
In article <4C493FB4.1050306@elvenkind.com>,
Taco Hoekwater
I am happy to announce a new *alpha* release of MetaPost:
---------------------------------------------------------------------- MetaPost 1.501 ----------------------------------------------------------------------
This is an *alpha* release following a major internal rewrite, so do not use this version for production purposes, but please test with as many files as you can.
I get errors when trying to use a mem file different from plain. The following: beginfig(0); path p; p := unitsquare xscaled 3cm yscaled 3cm; circular_shade(p, 0, yellow, red) ; endfig; end. compiled with mpost -mem=metafun test.mp results in a segmentation fault and an emtpy log file. And trying to typeset a MetaFont file with mpost -mem=mfplain test.mf produces the following error: This is MetaPost, version 1.501 (kpathsea version 6.0.0dev) 26 JUL 2010 09:45 **try-mfplain.mf (mfplain.mp Preloading the plain base, version 0.99: preliminaries, basic constants and mathematical macros, macros for converting units, macros and tables for various modes of operation, macros for drawing and filling, macros for proof labels and rules, macros for character and font administration, and a few last-minute items.) ! Attempt to close the bottom level file!. l.584 numeric mode,mag; % but leave mode and mag undefined I have just compiled mpost from source and moved it inside the TeX Live distribution. I have not re-built any format file as I have understood that they are not used any longer. Nicola
Hi Nicola, Thanks a lot for trying stuff! On 07/26/2010 09:48 AM, Nicola wrote:
I get errors when trying to use a mem file different from plain. The following:
beginfig(0); path p; p := unitsquare xscaled 3cm yscaled 3cm; circular_shade(p, 0, yellow, red) ; endfig; end.
compiled with
mpost -mem=metafun test.mp
results in a segmentation fault and an emtpy log file.
Same here, I'll try to fix that later today.
And trying to typeset a MetaFont file with
mpost -mem=mfplain test.mf
produces the following error:
This is MetaPost, version 1.501 (kpathsea version 6.0.0dev) 26 JUL 2010 09:45 **try-mfplain.mf (mfplain.mp Preloading the plain base, version 0.99: preliminaries, basic constants and mathematical macros, macros for converting units, macros and tables for various modes of operation, macros for drawing and filling, macros for proof labels and rules, macros for character and font administration, and a few last-minute items.) ! Attempt to close the bottom level file!.
This is not really a bug, more an incompatibility. In the new system, 'mem' files have to stop with 'dump', but the current mfplain.mp in the distrution doesn't do that yet (a similar problem is with plain.mp, but that is why we have mpost.mp). If you add a 'dump;' at the end of mfplain.mp, this problem will go away. Best wishes, Taco
In article <4C4D48C3.50803@elvenkind.com>,
Taco Hoekwater
Hi Nicola,
Thanks a lot for trying stuff!
On 07/26/2010 09:48 AM, Nicola wrote:
I get errors when trying to use a mem file different from plain.
And trying to typeset a MetaFont file with
mpost -mem=mfplain test.mf
produces the following error:
[snip]
This is not really a bug, more an incompatibility. In the new system, 'mem' files have to stop with 'dump', but the current mfplain.mp in the distrution doesn't do that yet (a similar problem is with plain.mp, but that is why we have mpost.mp).
If you add a 'dump;' at the end of mfplain.mp, this problem will go away.
Confirmed, that fixes the problem. One more problem I have with %& directives, as in this example: verbatimtex %&latex \documentclass{article} \begin{document} etex beginfig(1); label(btex x etex,cm*(1,1)); endfig; verbatimtex \end{document} \end etex end. It seems that '-tex latex' is mandatory in this case (%&latex is ignored). Is it by design? If so, I would like to point out that it may break code like the following: input TEX; TEXPRE("%&cont-en" & char(10) & "\starttext\setupbodyfont[8pt]"); TEXPOST("\stoptext"); verbatimtex %&latex \documentclass{minimal} \begin{document} etex beginfig(1); label(TEX("\type{This is} \ConTeXt"), (1cm,5mm)); draw btex \emph{And this is \LaTeX} etex; endfig; end. I know that this is borderline, but it compiles with mpost 1.2xx. Nicola
Hi, On 07/26/2010 11:34 AM, Nicola wrote:
I know that this is borderline, but it compiles with mpost 1.2xx.
Hm, it seems the problem is that the %&latex ends up on the second line of the output. This is caused by a purported bugfix patch that attempted to sanitize usage of verbatimtex (following the 'you should leave everthing inside verbatimtex...etex totally as is'-rule). From this example, it obvious that any whitespace starting before the TeX code in the first verbatimtex really does need to be stripped, because tex's --parse-first-line is not willing to look at the second line ;) This should not be a hard thing to fix (but the TeX interface is always tricky). Best wishes, Taco
In article <4C4D9E18.2000108@elvenkind.com>,
Taco Hoekwater
Hi,
On 07/26/2010 11:34 AM, Nicola wrote:
I know that this is borderline, but it compiles with mpost 1.2xx.
Hm, it seems the problem is that the %&latex ends up on the second line of the output. This is caused by a purported bugfix patch that attempted to sanitize usage of verbatimtex (following the 'you should leave everthing inside verbatimtex...etex totally as is'-rule).
From this example, it obvious that any whitespace starting before the TeX code in the first verbatimtex really does need to be stripped, because tex's --parse-first-line is not willing to look at the second line ;) This should not be a hard thing to fix (but the TeX interface is always tricky).
You're right (of course): using verbatimtex%&latex works. Nicola
participants (2)
-
Nicola -
Taco Hoekwater