Bo Peng wrote:
> < sorry for the PM to Leeming. >
Not a problem.
>> Loading an existing document containing such an inset results in no
>> previews being generated at all, as now.
>
> Locating a wrong formula among others is extremely difficult (may
> have to compile all formulas one by one.) so there is no good
> solution yet.
The information is present in the latex log file. I'm not saying that
it is easy to manipulate, but it is certainly possible.
Eg
ERROR_LINES=`sed -n '/^! Undefined control sequence.$/{
$!{
N
s/.*\n//
s/l\.\([^ ]*\) .*$/\1/
p
}
}
' ${logfile}`
Should extract '40' from:
! Undefined control sequence.
l.40 x=\alp
b\]
Thereafter, I'll leave it to you to exchange the previewed snippet
with your warning message. Line 40 here is 'x=\alp b\]'.
\begin{preview}
\[
x=\alp b\]
\end{preview}
Presumably, you'll start with
for line in $ERROR_LINES
do
...
done
>> In conclusion, the existiing solution and your proposal convey
>> appear equivalent information, but yours introduces different
>> (note, not 'better) behaviour for a 'just-edited' inset.
>
> If 'not-turning-to-image' is a definite and alarming sign of wrong
> formula, you are right. However, this is not the case.
>
> 1. Correct formula might not be previewed. One simplest example I
> can
> think of is: enter a \alpha, wait for preview, copy and paste
> this alpha. The new one will NOT be previewed unless you enter
> and leave it.
Then this seems to me to be a real bug.
> 2. The current indication of wrong formula is 'nothing happens to
> the
> formula afterwards'. This is psychologically hard to notice
> especially when lyx math box is a *normal* status of a formula.
> 3. previews need a while to appear, people will not wait and see the
> result. (Even when they wait, most likely: nothing happened
> yet... lyx is still compiling? instant-preview does not work?
> Just another not-shown?)
>
>> Forgive me, but I do not see why this is any improvement at all.
>
> I consider this patch as a minor improvement with minimal cost but
> of course you guys are the judges. :-)
Not at all. I'm quite happy to include such a a patch so long as it
doesn't increase the maintenance burden too much. However, I think it
should be more than just a gimmic, don't you? Ie, it's almost
certainly a good idea that it should do something useful when loading
a file too.
--
Angus
