Add liveliness bindings

[sashacmc]

Add the liveliness methods bindings:

• zc_liveliness_declaration_options_check
• zc_liveliness_declaration_options_drop
• zc_liveliness_declaration_options_null
• zc_liveliness_declare_subscriber
• zc_liveliness_declare_token
• zc_liveliness_get
• zc_liveliness_get_options_check
• zc_liveliness_get_options_default
• zc_liveliness_get_options_drop
• zc_liveliness_get_options_null
• zc_liveliness_subscriber_options_check
• zc_liveliness_subscriber_options_drop
• zc_liveliness_subscriber_options_null
• zc_liveliness_token_check
• zc_liveliness_token_null
• zc_liveliness_undeclare_token

Along with the examples:

• z_sub_liveliness
• z_get_liveliness

[Mallets]

Hello @sashacmc , in order to merge the PR we will need you to sign the Eclipse Contribution Agreement.

[Mallets]

The zenoh-commons.h header is meant to be common across zenoh-c and zenoh-pico.
All functions starting with zc_ are inherently meant to be specific to zenoh-c.
All these function declarations should hence go into zenoh-concrete.h.

@p-avital @milyin I see that other zc_ functions slipped in zenoh-commons.h in the past. I think they should be moved into zenoh-concrete.h as well. Opinions?

[p-avital]

Yeah, the original goal was to be able to copy-paste this header into pico to ensure we always have the same signatures, but it didn't end up working out that way.

I think we can make the splitter logic a bit better if we want to stick to the original ideal. Otherwise, we could remove the splitter entirely, or just leave it in that state. Any odd these solutions is fine by me :)

[Mallets]

zenoh-pico has evolved quite a bit since we initially made that split and I see how the split could hardly work out in practice. Therefore I'm more in favour of unifying all the includes in a single file rather than introducing artificial splits (although correct in principle).

For what concerns the scope of this specific PR, let's not change anything for the time being... A different PR may follow in the future.

[p-avital]

Agreed, though note that we'll never be single filed because the zenoh-macros file is still hand-written

[Yadunund]

fyi I ran into some issues when relying on z_check and z_drop macros with the new liveliness structs. Had to rely on the zc_liveliness_XX functions to check my tokens and undeclare them. More details: ros2/rmw_zenoh#67

[p-avital]

fyi I ran into some issues when relying on z_check and z_drop macros with the new liveliness structs. Had to rely on the zc_liveliness_XX functions to check my tokens and undeclare them. More details: ros2/rmw_zenoh#67

Is this the first time you've had issues with these macros? I seem to remember some incompatibility between C's generic macros and C++... The macros failing only on liveliness is strange considering the implementation of the macro is the same for them as for the rest...

[Yadunund]

fyi I ran into some issues when relying on z_check and z_drop macros with the new liveliness structs. Had to rely on the zc_liveliness_XX functions to check my tokens and undeclare them. More details: ros2/rmw_zenoh#67

Is this the first time you've had issues with these macros? I seem to remember some incompatibility between C's generic macros and C++... The macros failing only on liveliness is strange considering the implementation of the macro is the same for them as for the rest...

yup first time running into problems with the macros. I've been relying on them to check/drop owned sessions, publishers, subscribers without any problems.

[JEnoch]

@Yadunund the z_check macros for zc_owned_liveliness_token_t were indeed missing and are now available thanks to #195 .
I think the undefined symbol error calling z_drop() was probably an build environment issue.

I created ros2/rmw_zenoh#69 that hopefully will build successfully once workflows are approved.