Re: \ignoreprimitiveerror implemented

Hi Thanh, Karl

could I ask for a tiny change on this one perhaps?

Using the new feature (which we do) has resulted in a number of confused 
users as well as in problems in editors that look for errors in the log 
file.

Right now pdfTeX puts

    ignored error: Infinite glue shrinkage found in box being split

in the log, and while it says "ignored" people are worried that there is 
an error (which there isn't because it is deliberately ignored) and some 
tools that scan the log, return a run failure because they see the word 
"error".

I think all this could and should be easily avoided by altering the 
message string as suggested. Tools could be changed to behave but 
educating users that something  called an "error" is not is a rather 
hopeless task.

please consider making that change for 2026

thanks
frank


Am 25.01.25 um 23:23 schrieb Karl Berry:
> For the record: per Frank's request to Thanh (private mail) last July,
> Thanh has implemented a new \ignoreprimitiveerror bitwise
> parameter. It's enabled with -etex (which is set by default when
> building the formats, etc.)
> 
> If the value is 1, the \vsplit error
>    print_err("Infinite glue shrinkage found in box being split");
> is changed to a warning. All other values/bits are reserved (and
> currently do nothing).
> 
> I merged the change, using the same primitive name, into XeTeX. There
> appeared to be no easy way to make an engine-independent change file.
> I expect it is possible, but I gave up after a couple of failures.'
> Other things to do.
> 
> I asked Luigi(+ Hans) to implement it for LuaTeX and the JP devs to
> implement it for euptex. Those engines will probably have different
> interfaces. tbd. --thanks, karl.
>