KhronosGroup / Vulkan-Docs

The Vulkan API Specification and related tools
Other
2.83k stars 471 forks source link

`VkAccessFlagBits` broken link #1723

Open Dherse opened 2 years ago

Dherse commented 2 years ago

Hello,

In the documentation for VkAccessFlagBits there is a broken link: at line 1086 for synchronization.txt (note I'm using an older commit so this may be slightly incorrect), the link that gets generated with the text color, resolve, or depth/stencil resolve attachment gives the link #renderpass instead of https://www.khronos.org/registry/vulkan/specs/1.2-extensions/html/vkspec.html#renderpass.

Hope this is helpful,

Dherse

oddhack commented 2 years ago

Thanks. I'll take a look at it once things start up again after the holidays.

Dherse commented 2 years ago

Hello,

I am adding to this issue the following similar dead links:

Let me preface this by saying I don't know how Asciidoc works. But my guess looking at the code, is that when embedding pieces of doc from the spec into the individual doc items, it does not change the links to point to the spec.

Happy holidays,

Dherse

Dherse commented 2 years ago

Hello,

I found another example, in PFN_vkReallocationFunction.html and others, the link vkAllocationFunction_return_rules.html is broken.

There is also something very weird with the generation of PFN_vkInternalFreeNotification.html for the description section, where for some reason it generated a link in the description but puts it in a <code></code> block. The link cannot be clicked and has unusual text for the Vulkan docs.

Regards,

Dherse

oddhack commented 2 years ago

You're correct that the refpage extraction process doesn't always rewrite xrefs into links correctly. It does a pretty good job with the actual refpages blocks, with a few exceptions, but the common validity statements exist in separate asciidoc include files and xrefs there that should be rewritten, aren't. It is a known issue but not a high priority.

oddhack commented 2 years ago

N.b. what we'd probably have to do is change the refpage extraction scripts to actually insert the text of those common validity includes, rather than incorporate them via the 'include::' construct, prior to xref rewriting. It is not hard, but tedious.