eboasson

Absolutely. I have to say I find the current read/take interface quite illogical even for preallocated use: I'd rather provide a buffer of some size and have it use the...

> I've managed to get things working when only shared memory is used. Some bugs still remain, but the basic mechanism works. I'm facing an issue when I have a...

Thanks @sumanth-nirmal and @MatthiasKillat for checking this: > Therefore, if the sample cache is > 256 and samples arrive much faster than they are consumed we also may encounter this...

Hi @Zhang8912, that's because with TCP you need to know where to listen and where to connect to, and besides you can't have multiple TCP servers on a single port...

Hi @Zhang8912, assuming I understand you correctly, then I would say no, not really. TCP (and I mean TCP in general, not just the way it is used in DDS)...

@sgf201 I don't understand why you are pinging various people who aren't involved in this (nor why you didn't simply reopen #1097 after GitHub closed it automatically because I merged...

@sgf201 try this: https://github.com/eboasson/cyclonedds/commit/eff7e8208ea5c02dda72f206e5561748961b8334 If you call this new `dds_ruthless_reset_for_child_after_fork` function in the child after forking, it resets everything by just `memset`ting things. It worked for a simple experiment I...

Doing nothing means you, or anyone else in a similar situation, can't use Cyclone. This is such a simple solution that I don't see a problem supporting it. The documentation...

@TheFixer I'm sorry it is taking so long to review. It hasn't slipped through the cracks, it is just that it needs and deserves a careful look.

@TheFixer I am very sorry that it got ignored for so long. I don't know why I missed this: > @eboasson This PR implements delayed enabling of entities. A more...