Finding out the name of PL_compcv early in compilation

A question on inner details, for folks familiar with the internals of
op.c and friends...

After you have called newATTRSUB() or similar, the newly-created CV
most likely has either a CvNAME_HEK or a CvGV, on which you can find
out the CV's name. But this is only created and set by newATTRSUB()
itself, and isn't available at any earlier time during parsing the sub.
In particular, you can't find out the name of the sub while building
the optree or anything like that, when given only the value of
PL_compcv.

I notice that newATTRSUB() creates the name HEK or GV only from
information in the nameop, which was available in the parser much
earlier. In fact, init_named_cv() could have set it even.

Does anyone know why it is deferred so late? For my attributes-v2
branch, I would find it very useful to get access to the name HEK or GV
from out of PL_compcv, or at least by some other similar method of
access, when handling the early parse-time attributes.

I'm considering looking at just moving the code out of newATTRSUB() into
init_named_cv() to do this, but there's a lot of long twisty cases
involving trying to lazily reuse a GV or name HEK if an existing one is
there, and not break pointers here there and everywhere. It's a twisty
and subtle piece of logic, so it'll take me a while to really
understand why it's currently shaped like that.

Does anyone have any insight into it before I start?

-- 
Paul "LeoNerd" Evans

[email protected]
http://www.leonerd.org.uk/  |  https://metacpan.org/author/PEVANS