Dev command issues on v0.5.0

[hayk-skydio]

Describe the bug I'm having a series of build issues that are confusing me. Perhaps these should be split issues.

To Reproduce

1. Check out v0.5.0
2. Run rm -rf build && pip install -ve .
3. Verify that importing the library works, and the backend is symengine.

Issue 1: make lint and make check_types give an error:

gen/python/build/lib/sym/pose2.py: error: Duplicate module named "sym.pose2" (also at "gen/python/sym/pose2.py")
gen/python/build/lib/sym/pose2.py: note: Are you missing an __init__.py? Alternatively, consider using --exclude to avoid checking one of them.
Found 1 error in 1 file (errors prevented further checking)
make: *** [check_types] Error 2

Issue 2: make test_update does nothing:

make: Nothing to be done for `test_update'.

Issue 3: make docs fails:

File ~/miniconda3/envs/symforceX/lib/python3.10/site-packages/skymarshal/parser.py:290, in p_array_dim(p)
    286     p[0] = syntax_tree.ArrayDim(tuple())
    287 else:
    288     # Either 1 or 2 words in the brackets, passed through as a tuple.
    289     # May be a literal, a field name, an integer type, or a type and a virtual field name
--> 290     size_declaration = tuple(word for _, word in p[2:-1])
    291     p[0] = syntax_tree.ArrayDim(size_declaration)
    292     p[0].lineno = p[2][0]

File ~/miniconda3/envs/symforceX/lib/python3.10/site-packages/ply/yacc.py:198, in YaccProduction.__getitem__(self, n)
    197 def __getitem__(self,n):
--> 198     if n >= 0: return self.slice[n].value
    199     else: return self.stack[n].value

TypeError: '>=' not supported between instances of 'slice' and 'int'
TypeError: '>=' not supported between instances of 'slice' and 'int'

You can ignore this error by setting the following in conf.py:

    nbsphinx_allow_errors = True
make: *** [docs_html] Error 2

Environment (please complete the following information):

• OS and version: Darwin blitzer.local 21.3.0 Darwin Kernel Version 21.3.0: Wed Jan 5 21:37:58 PST 2022; root:xnu-8019.80.24~20/RELEASE_ARM64_T6000 arm64
• Python version Python 3.10.4
• SymForce Version v0.5.0

[hayk-skydio]

Issue 1 looks like it may have been fixed by @bradley-solliday-skydio in #189

[aaron-skydio]

Agree I think 1 should be fixed by #189 , and the other two should probably be their own issues

2. Hunch is that this is a "macOS sed doesn't match linux sed thing"? And that find test -name "*_codegen_test.py" | sed 's/test\/\(\w\+\).py/update_\1/g' is empty or something

3. This is seemingly a bug in ply that was fixed in 2013? Is your version of ply <= 3.8 somehow?

[bradley-solliday-skydio]

To summarize the state of this issue, issue 1 has been fixed for some time now, and issue 2 was resolved recently (here: https://github.com/symforce-org/symforce/pull/264). Aaron's hypothesis that this had to do with the different behavior of sed on macos was correct. Was resolved by updating sed's arguments.

Still need to try reproducing issue 3 with an old version of ply to confirm that that is the source of issue. If it is, I'll close the issue on the basis that I think the solution is to simply use a newer version of ply.

[aaron-skydio]

I don't think the ply thing is a symforce issue - if we decide it is because someone hits it again we'll make a separate issue