#7 "bug" in dvips destroys operation of klatexformula

[W Unruh]

There is a "bug" (feature) of dvips which can make klatexformula unuseable. You cannot use the t option ( pagesize option) either on the command line or in the texmf/dvips/config/config.ps file. klatexedit uses the dvips -E command to create and eps file which has a bounding box which is tightly constrained to the formula. But if the "t" option is used in the config.ps file, this destroys that ability, and when klatexformula runs gs it will take up to 110% cpu time for over 10 sec and not produce anything useful. You Must Not use the t option. (Or any of the other 10 options that stop dvips from creating eps files.

This is also a bug in klatexformula as it should see the error message from dvips "Can't create EPSF, Sorry" and warn the user that klatexformula is unuseable, and why.

The options that will destroy dvips -E are
-t option -- paper format.
-t landscape
manual feed
numcopies>1
collatedcopies>1
cropmarks
*iname==0
totalpages>1

The options liale to be used in the config.ps file are the t (papertype, eg a4, letter)
or manualfeed and the most likely is the t option.

Using the t option seems an easy way out of the problem that some distributions are from Europe and will have A4 as the default which you may
want to use the
t letter
option in config.ps, or vice versa in Europe. A better way is to make sure that the
@ letterSize 8.5in 11in
@+ ! %%DocumentPaperSizes: Letter

comes before the
@ A4size 594.99bp 841.99bp
@+ ! %%DocumentPaperSizes: a4
@+ %%PaperSize: A4

in config.ps (for letter size users, and vice versa for Europeans)

[Philippe Faist]

Well I don't think this is a bug of KLatexFormula, but rather a configuration error on the user's side. divps should be able to work with the -E option.

Is there a way in the command line to cancel the -t option specified in the config.ps file?

[W Unruh]

While the adding of the "t letter" option to config.ps is certainly a configuration option, (and I reported it here so that others could be made aware of it), the bug in klatexformula is in not reporting to the user that dvips does not institute the -E option, reports an error saying it is not able to, but klatexformula keeps plowing ahead as if that error never occured. This error makes klatexformula unuseable, not just because it then shows a full page image in its preview-- but because then gs takes 10-30 seconds to render the page, Ie, klatexformula should be on the lookout for that error, and report it back to the user so the user knows that there is a problem and where the problem resides. Otherwise they are liable to simply assume that klatexformula is a useless program and give up. It is not, it is a great and extremely useful program that any physicist who uses Linux (or I guess now Windows as well) would be grateful to have.

Finally it might be a good idea to install in the FAQ a brief description of the problem and how to fix it.
There is no way of canceling a "t " option in config.ps on the command line that I know of.

[Philippe Faist]

Ok I see now your point.

I think that this bug will be fixed in the new KLF 3.3 version as I have changed the procedure to transform the dvi into eps/pdf with a proper bounding box calculation with gs. Do you mind trying out the svn development version [1] and letting me know if the bug is still there?

Independently, I will consider adding a warning before rendering of a large area at high DPI (eg. a full page), so that klatexformula doesn't freeze.

[1] svn repository at https://klatexformula.svn.sourceforge.net/svnroot/klatexformula/trunk/klatexformula
viewvc web interface at http://klatexformula.svn.sourceforge.net/viewvc/klatexformula/trunk/klatexformula/