Boost logo

Boost :

From: Robert Ramey (ramey_at_[hidden])
Date: 2006-02-08 22:40:36


I'm sorry I haven't been able to get to this. I've just got to much stuff
going on.

In the past this issue has come up when extended_type_info was
registered in more than one DLL or in a DLL and in the executable.
In these cases it has been addressed by organizing the source code
so that serialization code is present in one and only one module.
This does take a little bit of care but does make the executable
as small as possible so its not a bad idea in any case. Reviewing
the demo_pimpl example might shed some light here as - even
though the its a library not a DLL its might help show what's
involved in making sure that serialization code is present
only in one module.

Another thing that might help would to trap on the extended_type_info
registration with the debugger. This should help show where
a duplicate registration might be getting called from.

Its been my intention to add a section to the manual along with
a test and demo to illustrate how to avoid problems in this area
but I haven't gotten to it. Also I haven't been successful in
figuring out how to that or bjam/test setup to test/demo my
example in tihs.

And I'm currently absorbed in a some higher priority (that is - paying)
projects.

Sorry I can' tbe of more help right now.

RObert Ramey

Marc VIALA wrote:
> Hi robert,
>
> We have been in touch about workaround or a better code organisation
> for avoiding any assert in extended_type_info.cpp @ line#71 if one
> tries to upload two different classes, each one implemented in its
> own DLL, see
> thread http://lists.boost.org/boost-users/2006/02/16839.php.
>
> During our discussion, you have mentioned a related "test", but I
> didn't
> find it within the boost distribution. So my questions are:
>
> 1. Could you give me the name of the test, the names of the test
> files;
> 2. Do you have tested my example, see the zip file, that demonstrates
> the assert;
> 3. Is it possible to comment the assert code in
> extended_type_info.cpp?
> 4. Do you have any other related links?
>
> Furthermore any complete recommendation would be welcomed because we
> are not able to progress in our developments without this issue.
>
> Thank you by advance to handle my request.
>
> PS: I didn't see your reply to my last message in the boost-user
> list. Did you miss it?
>
> Marc Viala
> _______________________________________________
> Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk