From: <lse...@ya...> - 2007-05-05 11:12:52
|
Dear MikTeXers, Consider the following file: \documentclass{report} \begin{document} Bad: \[ \sqrt{\frac{l}{1-\sin2\theta}}\neq \sqrt{\frac{l}{1-\cos2\theta}} \] \large Fine: \[ \sqrt{\frac{l}{1-\sin2\theta}}\neq \sqrt{\frac{l}{1-\cos2\theta}} \] \end{document} When processed by MikTeX 2.6 the output looks fine. But when processed by TeXlive, the output looks strange: the first two radicals are different in size, when they shouldn't (content is the same except for sin instead of cos). One should expect that the radicals come out at the same size, otherwise the result is rather unpleasing aesthetically (note that the problem only happens at 10pt, not at 12pt). This problem has been discussed at the spanish TeX list, and has leaked to comp.text.tex. There, someone suggested to compare logs from two different implementations (putting \showoutput before \begin{document}). I've done that. Running the file through MikTeX 2.6 and TeXlive 2005 (linux) and parsing diff's between logs yields: (MikTeX 2.6 first) 10c10 < ..\vbox(550.0+0.0)x345.0, glue set 461.9502fil --- > ..\vbox(550.0+0.0)x345.0, glue set 461.95018fil 13c13 < ...\hbox(6.94444+0.0)x345.0, glue set 310.27771fil --- > ...\hbox(6.94444+0.0)x345.0, glue set 310.27773fil 47c47 < .........\OMS/cmsy/m/n/10 --- > .........\OMS/cmsy/m/n/10 ^^@ 55c55 < .........\OML/cmm/m/it/10 --- > .........\OML/cmm/m/it/10 ^^R 83c83 < .........\OMS/cmsy/m/n/10 --- > .........\OMS/cmsy/m/n/10 ^^@ 91c91 < .........\OML/cmm/m/it/10 --- > .........\OML/cmm/m/it/10 ^^R 98c98 < ...\hbox(8.2+0.0)x345.0, glue set 303.40167fil --- > ...\hbox(8.2+0.0)x345.0, glue set 303.40166fil 133c133 < .........\OMS/cmsy/m/n/12 --- > .........\OMS/cmsy/m/n/12 ^^@ 141c141 < .........\OML/cmm/m/it/12 --- > .........\OML/cmm/m/it/12 ^^R 169c169 < .........\OMS/cmsy/m/n/12 --- > .........\OMS/cmsy/m/n/12 ^^@ 177c177 < .........\OML/cmm/m/it/12 --- > .........\OML/cmm/m/it/12 ^^R It looks like there are some rounding problems in one of the implementations. Some people think that this a grave bug in MikTeX 2.6 since all TeXlive since 10 years ago give the same result (the strange looking one) and reportedly also MikTeX 2.5. Could someone explain what is happening? Best regards, Luis Seidel --------------------------------- LLama Gratis a cualquier PC del Mundo. Llamadas a fijos y móviles desde 1 céntimo por minuto. http://es.voice.yahoo.com |