Le 15/12/2015 18:03, Richard Heck a écrit :
On 12/11/2015 05:36 PM, Guillaume Munch wrote: >> Dear list, >> >> >> Following the discussions in
<http://www.lyx.org/trac/ticket/9794>, >> here are a series of patches
meant to address #9794 (inset-modify >> tabular commands are incorrectly
disabled) and #4189 >> (Edit->Rows&Columns not shown when in mathed
insert). > > There is a somewhat more general issue here: Dispatched
LFUNs are > generally caught by the first inset that knows how to handle
the > relevant command. So this can cause problems with embedded insets.
I > did some work on this a while ago in the InsetModify-Incomplete >
branch of my personal repo: >
http://git.lyx.org/?p=developers/rgheck/lyx.git;a=summary I'm not > sure
how this relates to your patches. > > Richard >
Dear Richard
Please make sure that you reply to the list :)
Regarding 2.2.0:
It is clear that the change you are suggesting is bigger and requires
more testing. This does not look like something that can be done for
2.2.0. Thus my question is, how do we want to address #9794 for the next
two years? Because in any case, we need some new lfun which is not
AtPoint. Indeed, for tabular commands, it is meaningless to dispatch to
the inset in front of the cursor because we cannot guess which cell is
targeted.
This new lfun table-modify was the best option as far as the discussions
went in the ticket, and it does not preclude a bigger change for 2.3.0.
Therefore, can I please commit this patch?
Regarding 2.3.0:
You are right, while writing the patch I indeed thought that a bigger
problem was maybe that insets should not catch inset-modify commands not
meant for them. Similar issues remain in the tabular dialog for this
reason, as I wrote in the commit. I now think that one should, as you do
in your patch above, test for the type of inset passed to inset-modify
and leave the lfun undispatched when it is not the good one. This avoids
the problem mentioned by Jean-Marc in the above ticket regarding
"hardcoding knowledge of other insets everywhere". This would also
prevent issues like <http://www.lyx.org/trac/ticket/9386>, so much that
I wonder why this was not done already.
Sincerely
Guillaume