Add --force-output-type option

[dreifachstein]

Fixes #4841

[kripken]

This looks correct and also like a reasonable thing to add. But there was some debate on this in that issue, let's wait for feedback from @juj and others.

Also, this will need tests. Can add in tests/test_other.py, can search for similar emcc options and see how they are tested, for a basis to start from etc.

[dschuff]

+1, I think this is a good thing to have, and I also like this name for it, over the other ideas in #4841

[juj]

Make sure to change the merge target branch to incoming rather than master. Also, it is probably better to rebase to work on this feature on top of source incoming branch, since master is a bit of time away. (master=stable, incoming=latest dev)

[juj]

For this case in particular, it would be good to add a comment about what this does. I presume it will build to an object file? (same as .o?)

[dreifachstein]

Tests added.
Fix a case when specified js output filename contains no dots.

[klieth]

Any plans for merging this PR? I've run into the same issue as others in #4841 and would love to have this fix available.

[saschanaz]

Currently this PR seems to silently ignore --force-output-type when with -c, -E and -M. Shouldn't it be an error?

[stale]

This issue has been automatically marked as stale because there has been no activity in the past year. It will be closed automatically if no further activity occurs in the next 7 days. Feel free to re-open at any time if this issue is still relevant.

[RReverser]

Looks like this was accidentally abandoned; can we revisit / land it?

[sbc100]

I think this feature landed as --oformat= in #12640, unless I didn't understand fully this proposal?

[RReverser]

Huh, I'll have to check. I've grepped emcc --help for "out", didn't see anything relevant, and thought it's not there. In fact, it seems that oformat doesn't show up in --help or on https://emscripten.org/docs/tools_reference/emcc.html either.

[sbc100]

Ah, I guess i forgot to update the documentation. Sorry about that.

[RReverser]

No problem at all. But yeah, it would be helpful to add it.

Maybe emcc.py could use one of those argument parsers that auto-generate --help? Not sure if that's more effort or less though.

[sbc100]

No problem at all. But yeah, it would be helpful to add it.

Maybe emcc.py could use one of those argument parsers that auto-generate --help? Not sure if that's more effort or less though.

That would be great! IIRC i've tried it a few times in the past and always given up.. but I assume its not completely impossible.

[sbc100]

No problem at all. But yeah, it would be helpful to add it.
Maybe emcc.py could use one of those argument parsers that auto-generate --help? Not sure if that's more effort or less though.

That would be great! IIRC i've tried it a few times in the past and always given up.. but I assume its not completely impossible.

(One problem I seem to remember is that emcc passes any arguments it doesn't understand directly to clang.. also the ordering of flags and non-flags is import: e.g. -Wl,--whole-archive libfoo.a -Wl,--no-whole-archive. Its kind of tricky to not end up with something bespoke I think).

[RReverser]

(One problem I seem to remember is that emcc passes any arguments it doesn't understand directly to clang

Right, but emcc --help already mostly outputs its own options (not generic clang ones, except where meaning is overridden), so that wouldn't change much. Also, such auto-generated --help could always be complemented with manual options that need to be documented in addition to automatically parsed ones.

[sbc100]

I meant the use an auto-generated option parser (which comes with auto-generated help docs) its tricky in situations like emcc for various reasons. The main pitfalls/limitation that I remember are the two I mention above. Those make it hard to use optparse which is the normal python way to do such things.

I'm not saying its impossible, just that have tried and failed in the past.

[RReverser]

Yeah that's reasonable; just thought it might help to automate some work, but I recognise it might cause more problems than it solves.