Hi Olivier,
On Wednesday 01 February 2006 09:56, Olivier Monaco wrote:
> I tried to compile Gazebo with using :
> ./configure --enable-atrv --enable-b21r
>
> But configure ignore "enable-xxx" (Gazebo 0.5.2 and CVS version).
That's my fault - I fixed an earlier bug where configure interpreted
disable-xxx as enable-xxx, and inadvertantly made it ignore enable-xxx.
https://sf.net/mailarchive/message.php?msg_id=12974351
Earlier bug:
https://sf.net/mailarchive/message.php?msg_id=11160509
> I
> patched the "acinclude.m4" replacing "enable_name_lcase=no" line 21 by:
>
> ifelse($3,[yes],
> [enable_name_lcase=no],
> [enable_name_lcase=yes])
That fix has a slight quirk - it makes enable-xxx and disable-xxx have weird
semantics. Specifically, for a model which defaults to on (e.g. Blimp),
enable-blimp and disable-blimp both actually disable the model... Conversely,
for a default-off model (e.g. B21r), both enable-b21r and disable-b21r will
enable the model... (Unless I got mixed up... I did a quick test, but I don't
have my usual Gazebo dev environment handy just now...)
There are a couple of patches which retain the usual configure behaviour in
the bug tracker:
https://sf.net/tracker/?func=detail&atid=433164&aid=1293221&group_id=42445
I've described the patches in the email linked above, and both should be
effective - the only difference between them is in the ease of debugging the
resulting configure script. The longer "alternate" patch creates a configure
script which uses different variable names for the activation states of each
of the models, whereas the shorter three-line-patch re-uses the
"enable_name_lcase" variable as a temporary variable within each model
activation block (as is done at present).
Hope this helps,
Cheers,
Allwyn.
--
Allwyn Fernandes
Monash University
School of Applied Sciences and Engineering, Gippsland
Mobile: +61 414 470 392
|