Add libdev as a more powerful replacement for libtty#26614
Add libdev as a more powerful replacement for libtty#26614hoodmane wants to merge 10 commits intoemscripten-core:mainfrom
Conversation
sbc100
left a comment
sbc100
left a comment
There was a problem hiding this comment.
It would be good if you could add some documentation here.
Maybe in the PR description and/or at the top of the libdev.js file itserlf.
| self.assertContained('read file: /test.txt', output) | ||
|
|
||
| @crossplatform | ||
| @no_windows('opens /proc/self/fd/1') |
There was a problem hiding this comment.
But this test doesn't use NODERAWFS so the actual underlying FS should not effect the test, right?
There was a problem hiding this comment.
This PR changed the default behavior of the standard streams in Node so that isatty reflects the status of the Node streams. To test this, in test/other/test_node_term_size.mjs I close the original streams and reopen /dev/tty or /dev/null to force isatty to true/false. But then I need to put the output somewhere so I use open('/proc/self/fd/1') to dup the stdout file descriptor so the test can write the output.
There was a problem hiding this comment.
Oh I see. Don't you think the user should opt into that with either -sNODERAWFS of at the very least -sNODEFS? It seems like if neither of these options are set we should maybe not expose the real node FDs at all?
There was a problem hiding this comment.
Actually mabye you are right, it could be cool if std I/O streams just worked under node with no additional options (obviously if you do -sENVIRONMENT=web we would not expect any that to work).
There was a problem hiding this comment.
Does NODEFS have a way to take an already open Node file descriptor and turn it into a Emscripten descriptor? I didn't see a way to do that without manually creating a stream. I think it's fine to make a device for this -- it's quite similar to what stdio already does.
This is an automatic change generated by tools/maint/rebaseline_tests.py. The following (13) test expectation files were updated by running the tests with `--rebaseline`: ``` codesize/test_codesize_cxx_ctors1.json: 151832 => 154179 [+2347 bytes / +1.55%] codesize/test_codesize_cxx_ctors2.json: 151235 => 153582 [+2347 bytes / +1.55%] codesize/test_codesize_cxx_except.json: 195690 => 198039 [+2349 bytes / +1.20%] codesize/test_codesize_cxx_except_wasm.json: 166948 => 169296 [+2348 bytes / +1.41%] codesize/test_codesize_cxx_except_wasm_legacy.json: 164829 => 167177 [+2348 bytes / +1.42%] codesize/test_codesize_cxx_lto.json: 120519 => 122869 [+2350 bytes / +1.95%] codesize/test_codesize_cxx_mangle.json: 262181 => 264530 [+2349 bytes / +0.90%] codesize/test_codesize_cxx_noexcept.json: 153855 => 156202 [+2347 bytes / +1.53%] test/codesize/test_codesize_file_preload.expected.js updated codesize/test_codesize_file_preload.json: 23789 => 26139 [+2350 bytes / +9.88%] codesize/test_codesize_files_js_fs.json: 18215 => 20561 [+2346 bytes / +12.88%] codesize/test_codesize_hello_dylink.json: 43853 => 46204 [+2351 bytes / +5.36%] codesize/test_codesize_hello_dylink_all.json: 821790 => 824164 [+2374 bytes / +0.29%] Average change: +3.33% (+0.29% - +12.88%) ```
2 participants
WIP on #26612