Re: texi2any 7.0 performance regression (non-XS)

[pertusus]

On Sun, Feb 12, 2023 at 11:38:42AM +0000, Gavin Smith wrote:
> Indeed, the difference was at the first place I checked: on
> 2022-09-10, commit 46bfa18bb0, where the time came out as 43.9 sec.
> 
> 2022-09-10  Patrice Dumas  <pertusus@free.fr>
> 
>        * tp/Texinfo/ParserNonXS.pm (_parse_texi, _process_remaining_on_line):
>        add _process_remaining_on_line to match better with the XS parser.
> 
> 
> In 46bfa18bb0^ (the parent commit), the time came out at 26.0 sec, which isn't
> too bad.
> 
> I suspect it may just be that Perl subroutine calls are very slow, so it
> is faster to inline everything into one long subroutine that is thousands
> of lines long.
> 
> I'll try to look at this commit more to see if I can find anything else.

If I recall well, it is only rearranging of code to match better with
the XS parser, nothing else.  It is not the first time that this issue
of Perl subroutine calls being very slow pops up.  I prefer having a
code more similar to the XS parser than faster, by the way...

-- 
Pat