reduced Tokenizer::isCPP() usage / SymbolDatabase: removed mIsCpp and isCPP()
#5725
Conversation
firewave
changed the title
Tokenizer::isCPP()Tokenizer::isCPP() usage
firewave
added
the
merge-after-next-release
|
This might also speed up processing C code a bit as we previously forcefully used the C++ handling where it might not have been appropriate. We should merge this after the next release so the impact of the changes might be visible instead of getting lost in months of accumulated differences. |
|
The remaining usages which could be removed is in conjunction with types that wrap a token - and those are sparse. I also think we could switch from |
firewave
force-pushed
the
tokenizer-cpp
branch
from
6944e25
to
74fdbad
Compare
|
The tests which failed were added in ff9c04d. Looking at what they were testing for I wonder if these failures might be somewhat related to some regressions which occurred in 2.6: |
|
Further cleanups revealed potential issues/cases with undetermined standards. Those need to be resolved beforehand. Unfortunately that existing handling is quite awkward (surprise) so a fix is not straight forward and I currently can't wrap around it... It potentially might also require some other in-progress stuff to land first... |
firewave
removed
the
merge-after-next-release
firewave
force-pushed
the
tokenizer-cpp
branch
from
74fdbad
to
cb3a387
Compare
|
Requires #5853 to be merged first. |
firewave
force-pushed
the
tokenizer-cpp
branch
3 times, most recently
from
c488af5
to
e81a560
Compare
firewave
force-pushed
the
tokenizer-cpp
branch
from
e81a560
to
c725400
Compare
firewave
force-pushed
the
tokenizer-cpp
branch
4 times, most recently
from
947fb7a
to
cc41b98
Compare
lib/symboldatabase.cpp
Outdated
| @@ -53,7 +53,7 @@ | |||
| //--------------------------------------------------------------------------- | |||
|
|
|||
| SymbolDatabase::SymbolDatabase(Tokenizer& tokenizer, const Settings& settings, ErrorLogger* errorLogger) | |||
| : mTokenizer(tokenizer), mSettings(settings), mErrorLogger(errorLogger), mIsCpp(mTokenizer.isCPP()) | |||
There was a problem hiding this comment.
It seems like this could have been left in to avoid the repeated calls on the tokens but the language had not been determined at this point yet so this caused assertions in the unit tests.
I also think it is cleaner to have one less place which provides this information.
firewave
changed the title
Tokenizer::isCPP() usageTokenizer::isCPP() usage and some additional cleanups related to it
firewave
changed the title
Tokenizer::isCPP() usage and some additional cleanups related to itTokenizer::isCPP() usage / SymbolDatabase: removed mIsCpp and isCPP()
firewave
force-pushed
the
tokenizer-cpp
branch
3 times, most recently
from
c7d6290
to
80a2077
Compare
firewave
force-pushed
the
tokenizer-cpp
branch
2 times, most recently
from
c72a06b
to
2218041
Compare
…plicit-constructor` clang-tidy warning
…` clang-tidy and `functionStatic` selfcheck warning
Each
Token(should) be connected to aTokenList.Tokenizerjust encapsulates that so we have no need to check theTokenizerbut can simply ask theToken.Also if we have function calls we pass in a flag to tell it if it is C/C++ we can get rid of that flag and simply ask the
Token. In several cases we were only passing that flag to be used in some underlying calls making it completely unnecessary in the top call.