Received: by nummer-3.proteosys id <01C19443.992FD994@nummer-3.proteosys>; Thu, 3 Jan 2002 11:44:19 +0100 In-Reply-To: Your message of Thu, 30 Jan 92 18:07:53 +0100. <9201301725.AA26895@enet-gw.pa.dec.com> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C19443.992FD994" Return-Path: <@vm.gmd.de:LATEX-L@DHDURZ1.BITNET> X-MimeOLE: Produced By Microsoft Exchange V6.5 x-vm-v5-data: ([nil nil nil nil nil nil nil nil nil][nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil]) Content-class: urn:content-classes:message Subject: Re: Validating LaTeX output Date: Thu, 30 Jan 1992 19:40:38 +0100 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: From: "Leslie Lamport" Sender: "LaTeX-L Mailing list" To: "Rainer M. Schoepf" Reply-To: "LaTeX-L Mailing list" Status: R X-Status: X-Keywords: X-UID: 559 This is a multi-part message in MIME format. ------_=_NextPart_001_01C19443.992FD994 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable There are two purposes for a LaTeX "trip" file: 1. Regression testing of new releases, to make sure that fixing one bug doesn't break something else. 2. Providing document-style designers with a good input files for testing their new releases. I propose that regression testing be performed by diff'ing the .log files against their previous (presumably correct) versions. Note that: * Using \showoutput will put everything that's in the .dvi file in = the .log file. Careful design can keep the size of the resulting .log = files acceptable. ("A \hfill Z" is just as good as a full line of text for displaying the format.) * Most internal changes (as opposed to style changes) should have no effect on the output. Any changes caught by diff can be checked by visual inspection of the output. If the new version is judged to be acceptable, then it becomes the standard version for future = regression tests. The most difficult bugs to detect are those involving line and page breaks. LaTeX has some bugs that manifest themselves only when figures and/or footnotes and page breaks come at just the wrong place. Changing the position of something by a few points makes everything work fine. It will take some nasty "tripping" to test for this kind of bug. A special, very long trip file will probably be needed for checking the output routine. Leslie ------_=_NextPart_001_01C19443.992FD994 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Re: Validating LaTeX output

There are two purposes for a LaTeX "trip" = file:

   1. Regression testing of new releases, to = make sure that fixing one
      bug doesn't break = something else.

   2. Providing document-style designers = with a good input files for
      testing their new = releases.

I propose that regression testing be performed by = diff'ing the .log
files against their previous (presumably correct) = versions.  Note that:

   * Using \showoutput will put everything = that's in the .dvi file in the
   .log file.  Careful design can keep = the size of the resulting .log files
   acceptable.  ("A \hfill = Z" is just as good as a full line of text
   for displaying the format.)

   * Most internal changes (as opposed to = style changes) should have no
   effect on the output.  Any changes = caught by diff can be checked by
   visual inspection of the output.  = If the new version is judged to be
   acceptable, then it becomes the standard = version for future regression
   tests.

The most difficult bugs to detect are those involving = line and page
breaks.  LaTeX has some bugs that manifest = themselves only when figures
and/or footnotes and page breaks come at just the = wrong place.
Changing the position of something by a few points = makes everything
work fine.  It will take some nasty = "tripping" to test for this kind of
bug.  A special, very long trip file will = probably be needed for
checking the output routine.

Leslie

------_=_NextPart_001_01C19443.992FD994--