From mboxrd@z Thu Jan 1 00:00:00 1970 X-Msuck: nntp://news.gmane.io/gmane.comp.tex.context/2579 Path: main.gmane.org!not-for-mail From: Hans Hagen Newsgroups: gmane.comp.tex.context Subject: question concerning overful box warnings Date: Fri, 01 Sep 2000 18:08:25 +0200 Sender: owner-ntg-context@let.uu.nl Message-ID: <3.0.6.32.20000901180825.0088c8f0@pop.wxs.nl> NNTP-Posting-Host: coloc-standby.netfonds.no Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Trace: main.gmane.org 1035393363 10037 80.91.224.250 (23 Oct 2002 17:16:03 GMT) X-Complaints-To: usenet@main.gmane.org NNTP-Posting-Date: Wed, 23 Oct 2002 17:16:03 +0000 (UTC) Original-To: ntg-context@ntg.nl Xref: main.gmane.org gmane.comp.tex.context:2579 X-Report-Spam: http://spam.gmane.org/gmane.comp.tex.context:2579 Hi, As a result of some optimizations tex (that is the original one) can report confusing filenames when reporting overful boxes. If for instance fontx, fonty and fontz are all defined as 20pt the last defined name one is reported. This can lead to situations where an overful box is reported using a pseudo small caps, while actually another font switch is the cause. That way one can end up in endless lookups for the reason for the box being overful. A few years ago Taco and I spend some hours on the telephone hacking tex the program as well as low level context to find the reason for this behavior. We finally found that it's in tex the program, an optimization side effect. After that I decided that it was better to report a nonsence name than a wrong one, which is why since then context reported /font/ instead of /somename/. This was accomplished by a redundant definition of font/ with each font switch. Currently I'm implementing math time support (that is extending, upgrading taco's font-mty module) and at the same time I clean up the font mechanism a bit (since I'm also adding font size mapping, which is needed for math time, since in math e.g. 10.8 pt is used in math instead of 11pt; so in addition to "somefont sa 2" there will now be "somefont mo 2" too, in which case the mapped size will be taken if present). Now, while testing this, I ran into a strange phenomena. Suddenly instead of my /font/ messages I get font filenames like /tir/ and /tir@8.4pt/ [kind of funny that in the former case the design size is not reported]. A search on older machines learns me that at a certain moment something has been changed in tex the program (unless I'm wrong). Now, apart from the interesting question if this change is triptest compatible, if indeed it is a change, it demonstrates that even cosmetic changes can have strange consequences [let alone spoiling a few hours] so i wonder what happens on other machines/texs. The following test \dorecurse{400}{x} should give an overfull box with /something/..... mentioned as log. Now, what is it on other machines/tex's? Still /font/ or is there a name mentioned. I would appreciate if some users would test that (if possible on for instance pdftex13, miktex, etc). Hans [i wonder is there is a way to know such changes in advance, like the change in hyphen names of the dutch patterns, since i want context to be in tune with recent distributions] ------------------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com -------------------------------------------------------------------------