View Issue Details
ID | Project | Category | View Status | Date Submitted | Last Update |
---|---|---|---|---|---|
0000959 | luatex | luatex bug | public | 2015-12-09 01:51 | 2015-12-16 14:34 |
Reporter | karlberry | Assigned To | luigi scarso | ||
Priority | normal | Severity | major | Reproducibility | always |
Status | closed | Resolution | fixed | ||
Platform | i386-linux | OS | tex live | OS Version | 2015 |
Product Version | 0.80.0 | ||||
Summary | 0000959: bad dvi file made by dvilua(la)tex | ||||
Description | dvitype reports a bad dvi file after attached KERNEL.tex (latex) file is processed by dvilualatex. it seems this should never happen (obviously can't be affected at the macro level), so i hope you will consider debugging it as provided in the attached zip. it would take a huge effort to make a plain tex version. The other files in the zip (nma.cfg, main.mk4) are read during the tex run. | ||||
Steps To Reproduce | most conveniently, just "make" (in attached zip). it just runs dvilualatex and then dvitype to check the result, but the command line is so lengthy, I don't see the point of inserting it here. (The invocation comes from tex4ht.) The output I get ends with: ... Output written on KERNEL.dvi (1 page, 208428 bytes). Transcript written on KERNEL.log. dvitype KERNEL.dvi ... ' LuaTeX output 2015.12.08:1630' Bad DVI file: byte 208300 is not post! | ||||
Additional Information | The interest here is not in dvi output as such, but that tex4ht uses dvi as an intermediate format. The OP (Nasser M. Abbasi <nma@12000.org>) is using tex4ht extensively. It may not be possible to create a valid dvi file, I don't know, but in that case it seems that the engine should bail out with an error message instead of silently creating an erroneous file. The document itself has been cut down (believe it or not :), and any further removals, even of seemingly innocuous lines, eliminated the failure. All the document is doing is making a long "listing" of matlab code. In particular, commenting out the \usepackage of the ae fonts also makes the run succeed. Ok, there's no real reason to use the ae fonts, and there seems to be a related bug in tex4ht, but still, regardless of anything, the output should not be an invalid DVI file. This is with dvilua(la)tex from stock TL'15. Problems with tex4ht prevent running with earlier versions :(. Thanks. P.S. The field "Product Version" list appears strangely sorted, with 0.80.0 coming about halfway down, and lots of other "intermingling" ... | ||||
Tags | No tags attached. | ||||
|
|
|
Ok, thank you very much for the report. |
|
I think we have found the bug, and a fix. |
|
fixed rule function pointer (quite old bug probably) |
Date Modified | Username | Field | Change |
---|---|---|---|
2015-12-09 01:51 | karlberry | New Issue | |
2015-12-09 01:51 | karlberry | File Added: nma.zip | |
2015-12-09 08:38 | luigi scarso | Note Added: 0001562 | |
2015-12-11 11:25 | luigi scarso | Note Added: 0001563 | |
2015-12-16 14:29 | Hans Hagen | Assigned To | => luigi scarso |
2015-12-16 14:29 | Hans Hagen | Status | new => assigned |
2015-12-16 14:32 | Hans Hagen | Note Added: 0001565 | |
2015-12-16 14:32 | Hans Hagen | Status | assigned => resolved |
2015-12-16 14:32 | Hans Hagen | Fixed in Version | => 0.86 |
2015-12-16 14:32 | Hans Hagen | Resolution | open => fixed |
2015-12-16 14:34 | Hans Hagen | Status | resolved => closed |
2015-12-16 14:34 | Hans Hagen | Fixed in Version | 0.86 => |