# [tex-live] Location of recorder file

Philipp Stephani st_philipp at yahoo.de
Thu Apr 28 22:15:06 CEST 2011

```Am 28.04.2011 um 02:06 schrieb Karl Berry:

> Philipp,
>
> Getting back to the recorder file issues you wrote about starting
> with a message on 21 Sep 2010 16:28:58:
>
> - The only recent change was to insert the pid in the original fls name.
>  Ordinarily (in all engines), it still gets renamed to be the same as
>  the log file (s/log/fls).  This was done in response to a Debian bug
>  report: http://bugs.debian.org/575731.
>
> - All the other stray behavior you mention has been there "forever", as
>  far as I know.  TeX's command-line behavior is idiosyncratic to say
>  the best -- "texput", CTRL-D, etc., etc.  Once you start trying to
>  process TeX input on the command line, all bets on "normal" behavior
>  are off.
>
> - It's not possible for the fls file to always be the same as the log
>  file because the recording starts long before the log file is known.
>  Once you

That's fine as long as it is documented and there is a way around it (using --jobname). More important is the output directory issue, which is also present in all engines.

> - I have no good idea as to why the LaTeX-on-command-line example:
> pdflatex --recorder --output-directory=../b \
> '\documentclass{minimal}\csname @@end\endcsname'
>
>  ends up with the fls in the current directory, while the plain TeX
>  example puts it in the output-directory.  I can reproduce it on my
>  Unix system; also no idea why Windows is different, but not surprised.

In fact, plain LuaTeX also seems to behave wrong.

>  But the behavior is the same in older years, so I'm not going to
>  worry about it.  If you want to debug it and send a patch, great.

I'll try to investigate this on the LuaTeX side (with which I'm a bit more familiar).

> In lieu of code fixes, I wrote some caveats in web2c.texi regarding
> -recorder.  Sorry.