On Tue, 22 Jun 2021 09:41:20 +0300 Tzvetomir Stoyanov <tz.stoyanov@xxxxxxxxx> wrote: > I was missing that API when I worked on the host-guest time synch > algorithms, to initialize the library time synchronization context. I > decided to introduce a new API, for initializing only that part of the > library - tracecmd_tsync_init(), as this functionality will not be > used commonly. Now with the compression changes, there is a need for > an API for initializing the library compression context. As this I don't see that, but I'll reply to the other email on my thoughts there. > functionality will be used in almost all use cases, I think it makes > sense to introduce this global tracecmd_lib_init() API, instead of > something like tracecmd_lib_compression_init(). Personally I do not > like the approach to have init APIs for different parts of the > library, especially in case of widely used functionality. This will > complicate the usage and will force the users to know the library > internals. I would even suggest to remove the tracecmd_tsync_init() > API and to init time sync context in the tracecmd_lib_init(). The tsync_init() is a specific functionality and only for recording, but a lib_init() sounds very generic, and something that needs to be run if needed or not. I still don't see the need for it. -- Steve
|