linearize.h: sanitize header

Quoted 17 lines · Kamil Dudka

It's unfortunate to use 'true' and 'false' as identifiers in a system header. It clashes with corresponding macros from <stdbool.h> when included before <sparse/linearize.h>.

Signed-off-by Kamil Dudka <[email protected]>

---
 linearize.h |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

linearize.h +0 -0

index 2205082..50b3601 100644
@@ -328,7 +328,7 @@ struct entrypoint {
        struct instruction *entry;
 };
 

Quoted 4 lines

linearize.h +0 -1

-extern void insert_select(struct basic_block *bb, struct instruction
 *br, struct instruction *phi, pseudo_t true, pseudo_t false); +extern

void insert_select(struct basic_block *bb, struct instruction *br, struct instruction *phi, pseudo_t if_true, pseudo_t if_false);

I guess it is wise to change this in linearize.c as well. Mind sending a patch?

It's unfortunate to use 'true' and 'false' as identifiers in a system header. It clashes with corresponding macros from <stdbool.h> when included before <sparse/linearize.h>.

Signed-off-by Kamil Dudka <[email protected]>

---
 linearize.h |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

linearize.h +0 -0

index 2205082..50b3601 100644

Quoted 18 lines · On Thu August 6 2009 11:23:26 Hannes Eder wrote:

Quoted 17 lines · Kamil Dudka

It's unfortunate to use 'true' and 'false' as identifiers in a system header. It clashes with corresponding macros from <stdbool.h> when included before <sparse/linearize.h>.

Signed-off-by Kamil Dudka <[email protected]>

---
 linearize.h |    2 +-

 1 files changed, 1 insertions(+), 1 deletions(-)

linearize.h +0 -0

index 2205082..50b3601 100644
@@ -328,7 +328,7 @@ struct entrypoint {
         struct instruction *entry;
  };
 

Quoted 7 lines

Quoted 4 lines

linearize.h +0 -1

-extern void insert_select(struct basic_block *bb, struct instruction
 *br, struct instruction *phi, pseudo_t true, pseudo_t false); +extern

void insert_select(struct basic_block *bb, struct instruction *br, struct instruction *phi, pseudo_t if_true, pseudo_t if_false);

I guess it is wise to change this in linearize.c as well.  Mind sending a patch?

The question is if we need/want to :-) It's change of the working code for no real benefit. I am talking only about system-wide headers which can be included anywhere.

Quoted 6 lines

Quoted 2 lines

I guess it is wise to change this in linearize.c as well. Mind sending a patch?

The question is if we need/want to :-) It's change of the working code for no real benefit. I am talking only about system-wide headers which can be included anywhere.

Well I see at least one benefit, a small one though. Syntax highlighting is somewhat confused with "true" and "false", at least emacs is. They appear like the constants, where in fact they are variables.

The likelyhood to break the code by renaming this two variables is kinda low, no? And IHMO it was not so wise in the first place to pick these names. ;)

Quoted 11 lines · On Thu August 6 2009 11:39:11 Hannes Eder wrote:

Quoted 6 lines

Quoted 2 lines

I guess it is wise to change this in linearize.c as well.  Mind sending a patch?

The question is if we need/want to :-) It's change of the working code for no real benefit. I am talking only about system-wide headers which can be included anywhere.

Well I see at least one benefit, a small one though.  Syntax highlighting is somewhat confused with "true" and "false", at least emacs is.  They appear like the constants, where in fact they are variables.

I can confirm it's the same case with the vim's syntax highlighter.

Quoted 3 lines

The likelyhood to break the code by renaming this two variables is kinda low, no?  And IHMO it was not so wise in the first place to pick these names. ;)

I would contend that only two variables are affected. They are if we consider only headers. However the situation is much worse when we concern about .c files. The patch would be non-trivial. Please try the following command:

$ grep --color '[^_]false[^_]' *.c

Quoted 1 line

$ grep --color '[^_]false[^_]' *.c

$ grep --color '\bfalse\b\|\btrue\b' *.c | wc -l
91

some of them are just in comments, does not look to scary to me.  If others agree that its a good idea to rename them, I can do it if you don't want to.

I would just apply the change to the header file and related variables. The linearize.h is consider an API header file for other sparse application to use. So we'd better not assume too much on the sparse caller side.

I agree with Kamil that rename variable in linearize.c offer no real benefits. I consider it more of a personal preference thing. And it is internal to linearize.c. At this point renaming variable will mess up with annotations. It is not good enough reason to do it just to make the editor happy.

Quoted 9 lines · On Thursday 06 of August 2009 19:10:30 Christopher Li wrote:

I would just apply the change to the header file and related variables. The linearize.h is consider an API header file for other sparse application to use. So we'd better not assume too much on the sparse caller side.

I agree with Kamil that rename variable in linearize.c offer no real benefits. I consider it more of a personal preference thing. And it is internal to linearize.c. At this point renaming variable will mess up with annotations. It is not good enough reason to do it just to make the editor happy.

Well, let's make a tradeoff - we can only change the identifiers in linearize.h and the corresponding identifiers in linearize.c. I admit it could be confusing when we have different identifiers in the prototype and different identifiers in the function body. New version of the patch is attached.