Re: [bug #43122] texi2dvi does not compile enough times to get toc

[Gavin Smith]

On 18 September 2015 at 14:43, Gavin Smith <address@hidden> wrote:
> So this doesn't work. I reordered TEXINPUTS to put the working
> directory first, and the index displays as non-existent. I put in a
> breakpoint in texinfo.tex in \printindex (with \show), and had a look
> at the files in the directory. test.fls had already been created, but
> was empty.

Here's a plan.

* If \openout lines are output in the log files, use those by default.
This means that fl indexes will continue to work.
* Otherwise, use -recorder. Tweak the auxiliary file comparisons not
to check the *.fls file has changed, to prevent an infinite loop if
the file is overwritten by the recorder file.
* Change texinfo.tex not to output a file called *.fl - call it *._0
or use some other two-character suffix that contains non-alphabetic
characters.
* When the \openout lines are used, this change to texinfo.tex should
make no difference, and continue to work.
* Before this change to texinfo.tex, when the recorder is used, fl
indexes can end up blank. (Maybe texi2dvi could detect that
texinfo.tex output a *.fl file and issue a warning in that case.) That
would not be a regression, because this would only happen for TeX
variants like MikTeX where texi2dvi didn't run enough times in the
first place. For all other indexes, they would be displayed alright.
* After this change to texinfo.tex, fl indexes will be displayed when
-recorder is used.

I feel this plan could work and avoid any incompatibilities,
regressions, confusion, or inconvenience.