[Edbrowse-dev] ifndef CURLE_PEER_FAILED_VERIFICATION

[Edbrowse-dev] ifndef CURLE_PEER_FAILED_VERIFICATION

[Paul Onyschuk]

This part of code seems to be wrong in eb.h:

#ifndef CURLE_PEER_FAILED_VERIFICATION
#define CURLE_PEER_FAILED_VERIFICATION -1
#endif
#ifndef CURLE_REMOTE_ACCESS_DENIED
#define CURLE_REMOTE_ACCESS_DENIED -2
#endif

Since in curl/curl.h, those are enumerations, not macro constants:

typedef enum {
  CURLE_OK = 0,
  CURLE_REMOTE_ACCESS_DENIED,    /* 9 ...*/

Ifndef will be always true in this case, creating new macros. 

-- 
Paul Onyschuk

[Edbrowse-dev] ifndef CURLE_PEER_FAILED_VERIFICATION

[Karl Dahlke]

> let's remove them.

Done.

> I have a patch that eliminates a third of those clang warnings,

You can just push patches like these, I'm sure t'would be fine.

Karl Dahlke

Re: [Edbrowse-dev] ifndef CURLE_PEER_FAILED_VERIFICATION

[Chris Brannon]

Karl Dahlke <eklhad@comcast.net> writes:

> Chris I think you put those in, what do you think?

I honestly don't remember doing that, but I probably did, way back in
2008.  If there was a good reason for doing it back then, it's no longer
valid, so let's remove them.

On another note, I have a patch that eliminates a third of those clang
warnings, just by using some source annotations.  A bunch of warnings
are spurious, because clang has no way of proving that some functions
(showErrorAbort, i_printfExit) will never return.  Source annotations
help here.  I'm looking at the rest as well, and I'll make up a set of
several patches.

-- Chris

[Edbrowse-dev] ifndef CURLE_PEER_FAILED_VERIFICATION

[Karl Dahlke]

Your analysis is correct.
We are redefining these things to different numbers here - bad news.
The comment says these definitions are there for older
versions of curl that don't have them,
but honestly we don't support older versions of curl.
We gave up on that a while ago.
You just have to be up to date with curl, pcre, smjs, etc.
So I suggest we just delete this stuff.
I took it out and it builds and runs just fine.
Chris I think you put those in, what do you think?
If nobody objects I'll push the change and delete these lines
that, unintentionally, redefine two curl constants.

Karl Dahlke