>>>>> "Herbert" == Herbert Voss <[EMAIL PROTECTED]> writes:
Herbert> I have no problem with pendula.eps [I guess you mean pendula2.eps] You mean that convert pendula2.eps pendula2.png works, but convert pendula.epsi pendula.png does not work? What imagemagick version is that? Personnally, here is what I get: ----------- schuss: convert pendula.epsi pendula.png convert: Unable to open file (/tmp/magicqicNc7) [No such file or directory]. convert: Missing an image file name. schuss: convert pendula2.eps pendula2.png convert: Unable to open file (/tmp/magiceikz9h) [No such file or directory]. convert: Missing an image file name. schuss: convert -v|head Version: @(#)ImageMagick 5.2.7 01/01/01 Q:8 [EMAIL PROTECTED] Copyright: Copyright (C) 2001 ImageMagick Studio Usage: convert [options ...] file [ files... ] output_file Where options include: -adjoin join images into a single multi-image file -affine matrix drawing transform matrix -antialias remove pixel-aliasing -append append an image sequence schuss: --------------- If what you mean is that lyx works with it, yes, I know about that. Herbert> the attached file is an eps one, it has no preview part. I know... This is the pendula.epsi file that you posted in the message that I referenced, in which I removed the preview. My point is that the preview does not change anything to convert's problems, and thus epsi is irrelevant. Herbert> I modified some time ago filetools.C to detect eps-files with Herbert> a header like Herbert> %!PS-Adobe-2.0 EPSF-1.2 %%Title: (Flattened version of Herbert> pendula.ps) %%Creator: (PStill V1.5 (c) by Frank M. Siegert Herbert> 1997-2001) Herbert> as an epsi one so that my converter definition does the rest Where is this modification? I did not find it in the sources. Any way I'd rather not have it :) Herbert> standard epsi files work anyway. What do you mean? eps2eps is not an epsi->eps converter but a tool which completely rewrites an eps using gs pswrite driver. A side effect of that is that the epsi preview is gone. Big deal. My interpretation of the problem is that pendula.epsi did not suffer from being an epsi file, but from being a ps file (badly) wrapped into eps. If you type "gs pendula.epsi" from the command line, you will see that the gs prompt is "GS<1>" and not "GS>" as it is with any normal eps file. There is something bad going on. Christian Ridderström investigations suggest that the epsi file is somehow has several pages, which is of course a big no-no. So I guess we should just declare that pendula.epsi is broken So now we know that selecting on epsi is _wrong_, since epsi (ie the preview) is irrelevant here. You seem to propose to select separately files produced by pstill. But first you would have to prove that all files produced using this tool need some conversions. I have yet to see that (note that the bad properties of pendula.epsi are not due to pstill, but to the fact that it was a bad file that pstill did not manage to fix). Finally, you seem to think that running eps2eps to a file that does not need it (ie run it on _all_ epsi and pstill files although we have only one odd example of something going wrong) does not hurt. Unfortunately that is _wrong_ too: the file sent by Rob Lahaye (which is a `real' eps file) is completely killed by eps2eps creative rewrite (for some reason I do not understand). Try for yourself to run eps2eps on the attached eps file: it breaks the bounding box and turns the fonts into bitmaps... So, until someone tells me about a real use of epsi detection, I'm going to rip this off. Don't misunderstand me: I am ready to change my mind about it, but I need convincing facts. JMarc
X.eps.gz
Description: Binary data