[tlbuild] A strange probem with TL2016 mendex binary

[Nikola Lečić]

Hello all,

A TeX Live 2016 FreeBSD user from Japan sent me a report and analysis
of a strange behaviour of mendex binary. I'm not sure if it's
FreeBSD-specific, but... mendex tries to write to a log file with a
strange extension (I don't know it those symbols actually mean
something in Japanese).

Test files attached. If you run

platex -kanji=euc mendbug.tex
mendex -E -d mendbug.dic mendbug.idx

you will get something like

This is mendex version 2.6f [14-Aug-2009] (euc) (TeX Live 2016).
Scanning dictionary file ./mendbug.dic....done.
Scanning input file mendbug.idx.
Error: 愛上尾 is no entry in dictionary file in mendbug.idx, line
1....done (0 entries accepted, 1 rejected). 0 entries accepted, 1
rejected.
1 errors, written in mendbug.愛上尾.  
                     ^^^^^^^^^^^^^
Nothing written in output file.

Can someone shed some lite on what is going on here?

Excerpts from the anaysis:
-----------------------------------

  truss mendex -E -d mendbug.dic mendbug.idx

Then I got an evidence of strange character's as ...

     ...<snip>...
     mwrite(2,"m",1) = 1 (0x1)
     ewrite(2,"e",1) = 1 (0x1)
     nwrite(2,"n",1) = 1 (0x1)
     dwrite(2,"d",1) = 1 (0x1)
     bwrite(2,"b",1) = 1 (0x1)
     uwrite(2,"u",1) = 1 (0x1)
     gwrite(2,"g",1) = 1 (0x1)
     .write(2,".",1) = 1 (0x1)
==>  <B0>write(2,"\M-0",1) = 1 (0x1)
==>  <A6>write(2,"\M-&",1) = 1 (0x1)
==>  <BE>write(2,"\M->",1) = 1 (0x1)
==>  <E5>write(2,"\M-e",1) = 1 (0x1)
==>  <C8>write(2,"\M-H",1) = 1 (0x1)
==>  <F8>write(2,"\M-x",1) = 1 (0x1)  
     .write(2,".",1) = 1 (0x1)
     
     write(2,"\n",1) = 1 (0x1)
     Nwrite(2,"N",1) = 1 (0x1)
     owrite(2,"o",1) = 1 (0x1)
     twrite(2,"t",1) = 1 (0x1)
     hwrite(2,"h",1) = 1 (0x1)
     iwrite(2,"i",1) = 1 (0x1)
     nwrite(2,"n",1) = 1 (0x1)
     gwrite(2,"g",1) = 1 (0x1)
     write(2," ",1) = 1 (0x1)
     wwrite(2,"w",1) = 1 (0x1)
     ...<snip>...

I traced this table with gdb, if you like please check my trace report.


- - - < - - - - - - < - - - - - - < - - - - - - < - - - - - - < - - -  report START

$153 = 0x800e15218 "mendbug.ilg"
(gdb) n
431       cache (kpse, elt, ret);


(gdb) p logfile
$154 = 0x800e15218 “mendbug.ilg"

======= global valuable logfile still OK =====

(gdb) n
434       if (KPATHSEA_DEBUG_P (KPSE_DEBUG_EXPAND))

==== after above line 434 =====vvvvvvv 

(gdb) p logfile
$155 = 0x800e15218 "mendbug..”
====^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^=== BROKEN === suffice .ilg ==> .

(gdb) list
429       /* Remember the directory list we just found, in case future calls are
430          made with the same ELT.  */
431       cache (kpse, elt, ret);
432
433     #ifdef KPSE_DEBUG
434       if (KPATHSEA_DEBUG_P (KPSE_DEBUG_EXPAND))
435         {
436           DEBUGF1 ("path element %s =>", elt);
437           if (ret)
438             {

==== stack TRACE ====

(gdb) w
#0  kpathsea_element_dirs (kpse=0x642680 <kpse_def_inst>, elt=0x800e56600 ".") at ../../../texk/kpathsea/elt-dirs.c:434
#1  0x000000000041a5ad in path_search (kpse=0x642680 <kpse_def_inst>, path=0x420b9e ".", name=0x800e1cb40 "mendbug.dic", must_exist=1, all=0)
    at ../../../texk/kpathsea/pathsearch.c:291
#2  0x000000000041a7b2 in search (kpse=0x642680 <kpse_def_inst>, path=0x420b9e ".", original_name=0x800e170b0 "mendbug.dic", must_exist=1, all=0)
    at ../../../texk/kpathsea/pathsearch.c:376
#3  0x000000000041adb4 in kpathsea_path_search (kpse=0x642680 <kpse_def_inst>, path=0x420b9e ".", name=0x800e170b0 "mendbug.dic", must_exist=1)
    at ../../../texk/kpathsea/pathsearch.c:577
#4  0x000000000041ae3f in kpse_path_search (path=0x420b9e ".", name=0x800e170b0 "mendbug.dic", must_exist=1)
    at ../../../texk/kpathsea/pathsearch.c:598
#5  0x000000000040956d in KP_find_file (info=0x641dd0 <kp_dict>, name=0x800e170b0 "mendbug.dic") at ../../../texk/mendexk/kp.c:40
#6  0x0000000000401eb4 in dicread (filename=0x800e170b0 "mendbug.dic") at ../../../texk/mendexk/convert.c:74
#7  0x000000000040a46e in main (argc=5, argv=0x7fffffffea28) at ../../../texk/mendexk/main.c:303

- - - < - - - - - - < - - - - - - < - - - - - - < - - - - - - < - - -  report FINISH


-- 
Nikola Lečić = Никола Лечић
fingerprint : FEF3 66AF C90E EDC3 D878  7CDC 956D F4AB A377 1C9B
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 331 bytes
Desc: OpenPGP digital signature
URL: <http://tug.org/pipermail/tlbuild/attachments/20170404/6bab8758/attachment.sig>