[developers] [Fwd: Re: [spr30362] bug using climxm]

[Berthold Crysmann]

Berthold Crysmann wrote:
> Ann Copestake wrote:
>> I wonder whether this is really related to Andreas's patch - I've 
>> seen a similar error message in the past but never managed to obtain 
>> a reproducible situation.  In your case, does it stop the GUI working?
>>
>>
I've narrowed down the problem now: if I comment out (force-output t) in 
read-tdl-type-files-aux and check-generator-environment the error 
disappears.

Note also that the problem is not unique to the LKB toplevel window. 
Same happens, if I invoke these functions from the Emacs menu, so 
practically whenever we're not connected to some tty. Don't know though 
why this did not crop up before Andreas's patch....

Berthold



>>   
> Nope. It just aborts the function. Happens currently after loading a 
> grammar from the menu. Surprisingly enough, in this case the error 
> does not pop up immediately, but right after the first type file has 
> been read in (as far as I can tell from the LKB window). With 
> (index-for-generator) the error pops up straight away.
>
> Funnily enough, with Andreas's patch, or without emacs, nothing like 
> this happens. Likewise, with Andreas's patch in place and calling the 
> function from emacs (when output goes to emacs as well) is fine. The 
> only problem seems to be to call up the LKB with the Andreas's  patch 
> from emacs and invoking some commands from the menu like the two 
> above. Some menu commands where I do not observe this behaviour are 
> Parse|Parse Input..., but maybe the error message being sent out is a 
> different stream...
>
> I'll try as well with some other menu items that tend to produce a lot 
> of output in the LKB window. WIll tell you what came out of it.
>
> Berthold
>