You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Unfortunately, asciidoctor attributes are currently case-insensitive. This means that the asciidoctor enum attribute for CL_DEVICE_DEVICE_ENQUEUE_CAPABILITIES is indistinguishable for the asciidoctor type attribute for cl_device_device_enqueue_capabilities. When both attributes are indistinguishable, it looks like the last one "wins", and as a result there are several places in the spec where the type name is used in place of the enum name.
There is an Asciidoctor issue open to fix this, but it's been open for a while without any updates.
Options to fix this issue in the OpenCL specs:
Add a prefix or suffix to one class of attributes to disambiguate, probably the type name since there are fewer of them.
Special-case this particular attribute and try to avoid duplicates in the future.
I suppose we could also rename either the type or the enum to disambiguate - is there any other reason to prefer unique case-insensitive names? - but that feels a bit like the tail wagging the dog.
I'll probably bite the bullet and go with (1) if it doesn't look too bad.
The text was updated successfully, but these errors were encountered:
Unfortunately, asciidoctor attributes are currently case-insensitive. This means that the asciidoctor enum attribute for
CL_DEVICE_DEVICE_ENQUEUE_CAPABILITIES
is indistinguishable for the asciidoctor type attribute forcl_device_device_enqueue_capabilities
. When both attributes are indistinguishable, it looks like the last one "wins", and as a result there are several places in the spec where the type name is used in place of the enum name.There is an Asciidoctor issue open to fix this, but it's been open for a while without any updates.
Options to fix this issue in the OpenCL specs:
I suppose we could also rename either the type or the enum to disambiguate - is there any other reason to prefer unique case-insensitive names? - but that feels a bit like the tail wagging the dog.
I'll probably bite the bullet and go with (1) if it doesn't look too bad.
The text was updated successfully, but these errors were encountered: