Add tests for printf.jl

[thepulkitagarwal]

Since the error has to deal with the number of arguments, ArgumentError is more apt.

Also, in the same file, a similar statement (on Line 1167) is present:

throw(ArgumentError($macroname,": wrong number of arguments (",length(G),") should be (",$(length(sym_args)),")"))

[tkelman]

makes sense to me

[thepulkitagarwal]

#14410 might be relevant. I apologize for not seeing this earlier.

[tkelman]

It's okay. I think the author of that PR changed his mind about wanting to contribute, but if you'd like to take what he started with, rebase and clean it up for re-submission that would be welcome.

[thepulkitagarwal]

I'd like to do that. Thank you!

[thepulkitagarwal]

I added the tests, based on those in #14410.

[thepulkitagarwal]

I figured out the problem:

@sprintf("%p",42)
# returns 0x000000000000002a in a 64 bit system

which is 64 bits, but in a 32 bit system, it will return a 32 bit hex code. So, the tests will fail.

I checked the tests on integers and removed all tests that presumed 64 bits. But I need some help for floats. @tkelman Can you tell me what I should do to check that?

Thank you!

[StefanKarpinski]

Floats don't depend on system word size so those should be safe to leave unconditional. Like @tkelman suggested, the simplest way to fix the tests on 32-bit is to make the value you compare against conditionl on WORD_SIZE.

[thepulkitagarwal]

Okay! I have added the pointer tests based on the word size.

Also, I want to know what parts of code I should cover next. The manual method in #9493 for checking code coverage gives a very different result from Codecov. What should I do to get more accurate data?

[tkelman]

is num getting used anywhere?

[thepulkitagarwal]

It was supposed to be added in Line 30 (instead of 42). I'll add it now!

[thepulkitagarwal]

I have added a few more tests (two, to be exact) which were not covered earlier. I can rebase all the commits in this pull request into one commit before merging.

Also, is there anything else I should do?

Thank you!

[tkelman]

Nevermind my most recent comment which I deleted, I see there was a duplicated line in the existing test.

I think all this needs is a squash, nice work!

[thepulkitagarwal]

@tkelman I squashed all the commits. Can you check if everything's alright?

[tkelman]

Thanks!