Bunch of things undocumented

[zoffixznet]

If they are, I couldn't find where when searching the repo:

• [ ] syncpipe
• [ ] getcodelocation
• [ ] closefh_i
• [ ] queuepoll
• [x] getcurhllsym
• [x] bindhllsym
• [x] gethllsym
• [ ] const::PIPE_CAPTURE_ERR

[MasterDuke17]

https://gist.github.com/MasterDuke17/8833c501a6840ab9b2f29ea66d3db120 Here are all nqp::s where there's no entry for them in ops.markdown.

[coke]

FYI, there's a test for this from 2013 or so:

$ make
$ prove -v -e ./nqp t/docs/opcodes.t

This has 253 tests failing for undocumented opcodes, whereas the gist only has 206 - it reports on everything in the description, except the const:: entry.

[coke]

the 253 failures was for more than just undocc'd opcodes, and didn't include the js backend.

Latest totals:

17:32 < [Coke]> down to 387 failures in nqp's t/docs/opcodes.t 17:35 < [Coke]> of that, 258 are actual undocumented opcodes. 126 are opcodes that probably don't exist yet in js.

Note that the undocumented opcodes are across all three backends (and may suffer from bad guessing of what's an opcode)

[coke]

Down to 316 failures in nqp's t/docs/opcodes.t; Of that, 48 are opcodes that probably don't exist yet in js.

[coke]

Was a spike over 400 - with recent updates, down to 351 - they are all "missing documentation" errors, and note that there are several stubs in the ops documentation that should also be considered failures (but the test is only checking for headings, not actual docs)

[coke]

Down to 270 failures.

[coke]

260 failures. (Note that several of the passes are stubs with just the opcode name. :(

[coke]

Test no longer allows stubs with just the opcode name to pass, now reporting 289 failures.

[coke]

Snapshot of current failures at https://gist.github.com/coke/f87ab505c6cb825db51afcb5cb6a8990 (186 not ok)

[coke]

Updated the test file to skip moarvm opcodes that are found but not exposed. Fixed a few docs that had heading level docs to have per-opcode docs to satisfy the test. Updated the gist.

[coke]

back up to 617 failures earlier today due to changes in detection of ops in source file. committed a fix to the opcode test (b2880cb4c) back down to 191 failures. (with overlap: 160 missing on JVM, 125 JS, 163 moar)