Re: ConTeXt 2007.01.12 problem

[Sanjoy Mahajan <sanjoy@mrao.cam.ac.uk>]

> that test should go into the test repos 

I've committed it to the contexttest repo as
tex/context/base/supp-mis.tex/test-001.tex

By the way, in dvi mode the file fails unless "mpprocset:=0;" is put in the
\startMPpositiongraphic.  Here's the error msg:

  $ texexec --dvi --nonstopmode tex/context/base/supp-mis.tex/test-001.tex
  // on "TeX run 2"
  ...
  [MP as EPS ./to-dvi-mpgraph.4000] [MP color conversion ./to-dvi-mpgraph.4000
  ! Argument of \dodohandleMPcolor has an extra }.
  <inserted text> 
		  \par 
  <to be read again> 
		     }
  \includeMPasEPS ...\the \!!heightb }\doinsertfile 
						    }\wd \scratchbox \!!widt..

Taco had diagnosed this as due to ConTeXt not handling the new mpost
procsets (hence the mpprocset := 0;).  Not sure why that lack doesn't
hurt the run in pdf mode, but perhaps Taco or Hans understand the
reson.  The test file that I committed does not have the mpprocset:=0,
so it can be used to test for mpprocset handling in dvi mode.

The test skeleton script (./tex/context/base/enco-pfr.tex/Makefile,
which by the way I should move to a more general location) uses only
pdf mode.  I'll move it to a new home and modify it to test dvi mode
as well.

> (which btw we should get running now that it gets filled up); 
> i have to look into that asap

The ./tex/context/base/enco-pfr.tex/Makefile is a start.  But see a
previous email included below.  I still don't see where to get the
known-good comparison PDF (or ps).

-Sanjoy

`Not all those who wander are lost.' (J.R.R. Tolkien)


Date:    Fri, 06 Oct 2006 13:16:20 BST
From:    Sanjoy Mahajan <sanjoy@mrao.cam.ac.uk>
To:      contexttest-commits@foundry.supelec.fr
Reply-to: contexttest-commits@foundry.supelec.fr
Subject: [contexttest] automatic testing

[I added an actual subject line and started this msg as a new thread
-- sorry for breaking the threading and, in the first instance, for
lazily replying to the Cron msg rather than starting a new thread.]

Hans Hagen wrote:
  Sanjoy Mahajan wrote:
  > Just added a file to test \showframe (broken in the last few beta
  > releases).
  >
  > Which reminds me that I'm hoping someday to set up automatic testing
  > of new releases.  The main item in order to do that: 'known good'
  > outputs for the automatic comparison Makefile to use.  Is generating
  > each PDF with the latest non-beta release a good way?  It's probably
  > easier than looking at each one by hand, though not as reliable.
  >   
  we can think of several tests :

  (1) do they pass
  (2) does the file has the same number of pages
  (3) does a bitmap of the page roughly looks the same

  1/2 would already be great

I have 3 [bitmap checking] pretty much working (see the Makefile in
the repository).  It was a quick hack so is a bit rigid: It tells you
how different the old and new bitmaps are, but a shift of one pixel is
interpreted as "very different" (it doesn't do any dynamic programming
to do the match, just a simple overlay).

For 1 [do they pass], do you mean whether the texexec completes?  That
should be easy as well.

#2 [number of pages] should also be easy, and can be added to #3
pretty easily.  But I still want to know "same number of pages as
which release?"  Because all one has in the repository is .tex source
files, but not known correct PDF or log output from a correct run to
compare against.

Another question is what to do about settings like \runMPgraphics or
\runMPTEXgraphics (and write18).  For example, static MP figures break
(I'm pretty sure) if write18 is on but those other two are off.
Should the test files themselves turn on those variables if they are
needed (i.e. make any changes relative to the default cont-sys.rme)?

[I'll soon visit relatives for the upcoming U.S. long weekend so might
not reply for a few days, depends on how lazy I feel about email.]

-Sanjoy

`Never underestimate the evil of which men of power are capable.'
         --Bertrand Russell, _War Crimes in Vietnam_, chapter 1.
_______________________________________________
Contexttest-commits mailing list
Contexttest-commits@foundry.supelec.fr
https://foundry.supelec.fr/mailman/listinfo/contexttest-commits