Am 17.04.2013 13:03, schrieb Jean-Marc Lasgouttes:

Please find enclosed a patch that should fix the remaining problems (apply on 
top of the previous
one). I refrained from rewriting large parts of the code due to your comment 
below on doing too
complicated things. There is code duplication in GuiBox and the logic is often 
very difficult to
understand.
I will commit the pair of patches if they work for you.

I found now the time to test and your patch does not help. One still gets "0pt" as length. You can test this by inserting a box, setting its inner box to makebox and not checking the width and eventually pressing Apply.

In the View source window you must in this case see \mbox{} if it works.

My solution does indeed not look nice but is with a closer look simpler
than all others that we found so far. So is it really worth it to
implement a new empty length feature? I mean that is much more code than
having in 2 files a special code for en empty length?

The work I did on empty length will allow to cleanup other code (think
about the ugly support for box height, for example?). I want all these tests 
for special values of
lengths to just disappear, because they create a lot of maintenance burden.

I fully support you here but cannot help you as I don't understand the background. The problem is that your attempts don't work and I am lost here.

Do you really think that the code in InsetBox/GuiBox is not cleaner after my 
patch?

If it does work, then yes ;-)

thanks and regards
Uwe

Reply via email to