serialization of variant

Hi! I was working with an older version of boost which didn't have support for serializing variants. So I started to implement serialization of variants myself. When I looked into a newer version of boost I found serialization/variant.hpp . So I compared the boost version with mine. I found that I am missing some things, but I implemented the type lookup with a plain switch instead of repeating if-statements in a recursive function. I want to contribute my switch-code to boost but I don't know whom to email my code. Frank

Try attaching it to your response. Robert Ramey Frank Birbacher wrote:
Hi!
I was working with an older version of boost which didn't have support for serializing variants. So I started to implement serialization of variants myself.
When I looked into a newer version of boost I found serialization/variant.hpp . So I compared the boost version with mine. I found that I am missing some things, but I implemented the type lookup with a plain switch instead of repeating if-statements in a recursive function.
I want to contribute my switch-code to boost but I don't know whom to email my code.
Frank
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Hi! Well now, any comments? Shall I merge the existing variant.hpp and mine and then post again? I'm mostly concerned about saving "which", which is a stack variable. Is the address of this variable tracked while serializing? When saving multiple variants is there a chance that any occurrence of "which" is not saved, because the variable has the same address than some instance before already had? Thanks, Frank

Sorry, I haven't been able to study it with the attention it deserves. How would it be an improvement over the current implementation in CVS which is has been proven to pass all versions except borland? Robert Ramey Frank Birbacher wrote:
Hi!
Well now, any comments? Shall I merge the existing variant.hpp and mine and then post again?
I'm mostly concerned about saving "which", which is a stack variable. Is the address of this variable tracked while serializing? When saving multiple variants is there a chance that any occurrence of "which" is not saved, because the variable has the same address than some instance before already had?
Thanks, Frank
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Hi! Robert Ramey wrote:
Sorry, I haven't been able to study it with the attention it deserves. How would it be an improvement over the current implementation in CVS which is has been proven to pass all versions except borland?
The improvement is the use of "switch" instead of "if". I'll post a more mature version of my idea soon. Frank
participants (2)
-
Frank Birbacher
-
Robert Ramey