Still for productivity's sake, output is better than no output. This is especially important for large works, where the author does not want complete failure because of some inconsistency.
I am suggesting that the error message still shows up in the output. I suggest a variable that can be used for strict checking. > Sent: Saturday, November 14, 2020 at 4:09 PM > From: "Patrice Dumas" <[email protected]> > To: "Christopher Dimech" <[email protected]> > Cc: "help-texinfo gnu" <[email protected]>, "Gavin Smith" > <[email protected]> > Subject: Re: Failure in html output - Reference to nonexistent nodes > > On Sat, Nov 14, 2020 at 03:13:27PM +0100, Christopher Dimech wrote: > > Some time ago I discussed the problem that calling a reference > > to a nonexistent node when generating .info files. > > > > When outputting html, reference to nonexistent node also gives a > > problem. > > > > @ref reference to nonexistent node `igc03--Sicl--CircAmRg' > > That seems to be a good thing to me, this is an clearly an error. > > > Problem is aggravated because the failure stops the > > generation of any files. > > > > When using texi2pdf, one gets in the output pdf document > > > > <undefined> [igc03–Sicl–CircAmRg], page <undefined>, > > > > This is much better than not generating any output. > > My wild guess is that opinions are probably divided on that matter. > > But there is a possibility with texi2any to generate input even in case > of errors, with the option --force. The option --error-limit may also be > of use in that context. There is also --no-warn that an be used to > silence the warnings. > > -- > Pat >
