Recently I reported 2 problems related to metapost usage: 1. the loss of transparent color handling 2. textext not being able to handle %'s as in textext("15\%") The source of both problems can be nailed down to the one change in supp-mps.tex: \ifx\TEXEXECcommand \undefined \def\TEXEXECcommand{texmfstart texexec} \fi If I understand correctly, this effectuates a change from the perl to the ruby scripts. Removing the texmfstart solves the problem. I compared two runs for the transparency problem in order to see if there is some clue: 1. Both mp-files turn out being the same (except for the random seed). 2. The mpgraph.1 files differ only slightly, except for a few roundoff differences: new format extra lines: (but not a problem, I would guess) %%MetaPostSpecials: 2.0 123 1000 %%HiResBoundingBox: -5.66927 -5.66927 359.9991 119.05481 %%MetaPostSpecial: 7 1 1 1 1 0.94118 1 3 I had in my .bashrc: CONTEXT_RUBYSCRIPTS=/usr/local/teTeX/share/texmf-local/scripts/ context/ruby/ export CONTEXT_RUBYSCRIPTS and had CONTEXT_RUBYSCRIPTS in my path. Carefully had the executable and read-allowance bits set on all the ruby scripts. I guess it must have something to do with the ruby scripts, probably MacOSX unique, I fear. Something done differently with regard to the perl scripts perhaps? Taco already hinted at the possibility of some end-of-line problem. Alas, my knowledge of ruby is nonexistent, so I have no idea what and where to look for. I tried hard to find a solution, but I really have no firm idea what could cause these problems. Hans van der Meer
Hans van der Meer wrote:
2. The mpgraph.1 files differ only slightly, except for a few roundoff differences: new format extra lines: (but not a problem, I would guess) %%MetaPostSpecials: 2.0 123 1000 %%HiResBoundingBox: -5.66927 -5.66927 359.9991 119.05481 %%MetaPostSpecial: 7 1 1 1 1 0.94118 1 3
Believe it or not, but this is a known problem within web2c. It is triggered by an interaction between the new texexec and having two different entries in texmf.cnf for Metaposts memory size, for example like this: main_memory.mpost = 500000 main_memory.metafun = 3000000 Now, the metafun format is generated as "mpost", but the graphics are created as "metafun", and the different memory sizes make the specials go disappear. Strangely, I thought Hans and I had fixed by that a few weeks back by changing the mpost commandlines, but apparently we didn't. A workaround (while we sort this out) is to make sure that the .mpost and .metafun settings use the same (highest) number. Taco
On Jun 14, 2006, at 12:35, Hans van der Meer wrote:
Recently I reported 2 problems related to metapost usage: 1. the loss of transparent color handling 2. textext not being able to handle %'s as in textext("15\%")
The first problem seems to be solved by Taco's remark:
Believe it or not, but this is a known problem within web2c. It is triggered by an interaction between the new texexec and having two different entries in texmf.cnf for Metaposts memory size, for example like this:
main_memory.mpost = 500000 main_memory.metafun = 3000000
Now, the metafun format is generated as "mpost", but the graphics are created as "metafun", and the different memory sizes make the specials go disappear.
I checked my texmf.cnf in teTeX's main directory ../teTeX/share/texmf (the one I use) and to my surprise found the complete absence there of main_memory.metafun although context itself is present. Addding and regenerating metafun repared the "broken transparency" problem. I have further to investigate why apparently context.cnf seems not to have been read or being effective, because there main_memory.metafun is present! Thanks Taco for your remark, although it took some time before I realised that this remark of yours could be the key to the problem. However the breaking of the handling of "\%"'s in textext was not solved by this repair. Will there be a fix in the forseeable future? Taco wrote in reply to my first signalling the problem:
Hans van der Meer wrote:
I am afraid a serious error has been introduced into textext in the last update(s).
This used to work in metapost for typesetting text with a %-sign:
label(textext("66\%"), origin);
Now the % somewhere ends a line prematurely because the job aborts with: ! mpx file ended unexpectedly.
Can Hans please repair this, suddenly a lot of my typesetting is broken.
Verified.
Thanks for analizing this, I was wondering where those ! mpx file ended unexpectedly. messages came from in a document Mojca sent me.
Taco
Hans van der Meer
Hans van der Meer wrote:
However the breaking of the handling of "\%"'s in textext was not solved by this repair. Will there be a fix in the forseeable future? i'll check it, for the moment use: \percent\
----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
participants (3)
-
Hans Hagen -
Hans van der Meer -
Taco Hoekwater