Change m_bytes to unsigned in FixedBytesType

[jdlee23]

Fixes #3965.
Closes #4036.
Closes #4010.

🙏 sorry i left a mess on that other PR.

any chance i could poke one of yall to help me set up the local testing environment? realized changing numBytes might lead me down a bit of a rabbit hole, and i didn't want to continuously push! i'll buy yall a round of beers

[axic]

You can also click on the Circleci log and find out the reason for failures:

/root/project/libsolidity/ast/Types.cpp: In member function 'virtual bool dev::solidity::RationalNumberType::isImplicitlyConvertibleTo(const dev::solidity::Type&) const':
/root/project/libsolidity/ast/Types.cpp:876:38: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]
     return fixedBytes.numBytes() * 8 >= integerType()->numBits();
            ~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~

Please adjust the code for it to compile.

[jdlee23]

cheers @axic

I updated the PR to fix any compilation errors. I don't mind not receiving any bounty for this, although I'd love to contribute more slowly on the easier tickets.

[axic]

Please rebase this over #4036.

[axic]

Why did you chose to make totalBits unsigned while leaving fractionalDigits signed?

[jdlee23]

I just changed everything that m_bytes was connected to from int to unsigned along with warning/errors until the compiler was happy. Should I update _factionalDigits to unsigned as well?

[axic]

If you rebase it over #4036 then you don't need to.

[chriseth]

There is another int in CompilerUtils.cpp:691 and CompilerUtils.cpp:799, but then I think that's almost it!

Test errors.

[chriseth]

Sorry, but there is at least one compiler error:

/root/project/libsolidity/ast/Types.cpp:603:5: error: comparison of unsigned expression >= 0 is always true [-Werror=type-limits]
   0 <= m_fractionalDigits && m_fractionalDigits <= 80,
   ~~^~~~

[jdlee23]

nice catch

[chriseth]

Hm, I'm not sure if we actually force m_fractionalDigits to be nonzero somewhere.

[jdlee23]

Should I keep it as an int?

[chriseth]

No, just remove the lower bound check.

[chriseth]

Please incorporate the last commit into the appropriate bigger one.

[jdlee23]

@axic does this look good?

[chriseth]

We have a weird test failure:

ASSERTION FAILURE:
- file   : SolidityEndToEndTest.cpp
- line   : 3863
- message: Invalid encoded data
   Result                                                           Expectation
 X ................................................................ .........................da99.5c69723ff2b35428c8.cde458eff965dee

restarted the test.

[jdlee23]

thanks @chriseth !