Add a check for boolean expressions containing pointless/overriding connectives about lint HOT 5 CLOSED

It is indeed a better fit for lint than vet.

I'm not sure about your first four (comparing against true/false). It's true that they are equivalent, but code can carry semantic emphasis by doing that, so I'm wary of flagging them as wrong.

The remainder, now that I think about it, might match vet better. Though they are not incorrect code, they may be accidentally hiding code, and flagging them there probably makes sense in the same way that flagging unreachable code makes sense. I wonder what Rob thinks about that idea.

from lint.

I'm not sure about your first four (comparing against true/false). It's true that they are equivalent, but code can carry semantic emphasis by doing that, so I'm wary of flagging them as wrong.

Do you have an example handy?

Flipping through instances from the corpus, comparing against true/false appears to be a habit of the developer, rather than being reserved for emphasis.

The remainder, now that I think about it, might match vet better. Though they are not incorrect code, they may be accidentally hiding code, and flagging them there probably makes sense in the same way that flagging unreachable code makes sense.

That was the theory I held when I wrote it as a vet check, but it didn't survive first contact. I did some sampling, and I did not find any flagged code that was clearly buggy. I found lots of code that I didn't understand -- and which thus might have been buggy -- but nothing obviously incorrect.

I'd prefer to be wrong about this though, so feel free to push on it. :)

from lint.

On 28 June 2014 01:53, Josh Bleecher Snyder [email protected] wrote:

Do you have an example handy?

I don't have an example handy, but it reminds me of test code for
functions that return a bool, and sometimes I find it clearer to write
if foo(tt.in) == false. I don't think that's necessarily that much
worse than if !foo(tt.in) to warrant warning people about it.

That was the theory I held when I wrote it as a vet check, but it didn't survive first contact. I did some sampling, and I did not find any flagged code that was clearly buggy. I found lots of code that I didn't understand -- and which thus might have been buggy -- but nothing obviously incorrect.

I'd prefer to be wrong about this though, so feel free to push on it. :)

If it empirically doesn't find any bad code, or very little bad code,
we should probably just drop it.

from lint.

Sounds right. Dropped! Thanks for the helpful discussion.

from lint.

m

from lint.