Insufficient documentation for sfVulkan_getGraphicsRequiredInstanceExtensions?

[mgrojo]

I'm working on updating the Ada binding to version 2.6.0. There's a particular function that I don't know how to interface to. Forgive me if it's due to a lack of C/C++ knowledge on my part, but in any case, better documentation on how to use this function would be appreciated:

////////////////////////////////////////////////////////////
/// \brief Get Vulkan instance extensions required for graphics
///
/// \return Vulkan instance extensions required for graphics
///
////////////////////////////////////////////////////////////
CSFML_WINDOW_API const char* const* sfVulkan_getGraphicsRequiredInstanceExtensions(void);

I suppose that it returns a list of strings, but how is the caller aware of how many of them are available? For v2.5 I had to interface to a similar case, and the number of strings was returned through a count parameter. Compare with this case:

////////////////////////////////////////////////////////////
/// \brief Get a list of the names of all availabe audio capture devices
///
/// This function returns an array of strings (null terminated),
/// containing the names of all availabe audio capture devices.
/// If no devices are available then NULL is returned.
///
/// \param count Pointer to a variable that will be filled with the number of modes in the array
///
/// \return An array of strings containing the names
///
////////////////////////////////////////////////////////////
CSFML_AUDIO_API const char** sfSoundRecorder_getAvailableDevices(size_t* count);

[kimci86]

It looks like an oversight in CSFML API rather than a documentation problem.
The function returning a pointer is not enough. There is no way to get the number of items.

[ChrisThrasher]

That’s plausibly an oversight on my part. Not sure how to fix this without an API break but I’m not necessarily opposed to doing that. The API in its current form isn’t particularly useable so what does it matter to break usages of an unusable API?

Damned C APIs…

[mgrojo]

I guess I'm the first one trying to use it. If there are no current users of this Vulkan API, there's no reason to keep it as is; it'd better to fix it and release it as soon as possible.

[eXpl0it3r]

That’s plausibly an oversight on my part. Not sure how to fix this without an API break

You're not really breaking APIs when fixing broken APIs, at least in my book 😄