RE: devchk ideas
"Banginwar, Rajesh" <[email protected]> Thu, 27 Oct 2005 09:42:29 -0700
| Newsgroups | gmane.linux.lsb.test-suite |
|---|---|
| Message-ID | <B5677069E3D9994D9EE1C7295072EA960809107E@orsmsx402.amr.corp.intel.com> |
No, this will work only with functions and global variables. For macros, we will have to come up with different strategy. I believe except for integer, float or string macros, not much is tested in devchk. Many of the macros are marked as unknown for this purpose. I have one suggestion for this: For macros like #define foo bar we can introduce a new type string2 which will allow us to do string comparison. This is in addition to string type for macros like #define foo "bar". Thanks, -Rajesh > -----Original Message----- > From: [email protected] [mailto:lsb-test- > [email protected]] On Behalf Of ShrikantGangoda > Sent: Thursday, October 27, 2005 6:01 AM > To: [email protected] > Subject: [Lsb-test] devchk ideas > > Hi Rajesh, > I am working on devchk with pradosh. > > On Tue, 2005-10-25 at 11:24 +0530, Pradosh Adoni wrote: > > -------- Forwarded Message -------- > > > > > > Date: Mon, 24 Oct 2005 09:40:48 -0700 > > > > > > Hi Pradosh, > > > There is no transcript as such; it was just one liner from Nick. > > > > > > The idea is to use gcc extension to C: typeof. Declare two variables, > > > one with typeof (&function_name), and the other using the prototype > info > > > from DB. If you can assign one variable to the other without warning, > > > they have the same type. > > > > I have implemented the idea to check two functions.But,this idea can > not be used in checking macros as types are not define. > Example: #define ABC(a,b) (a+b) > here we don't have information on types of 'a' and 'b'. > > > > > > Hope this helps. This will be compile time check and we will have to > > > monitor warnings, may be there is a flag to gcc which can consider > > > warnings as errors. > > > > > > - Shrikant > > _______________________________________________ > lsb-test mailing list > [email protected] > http://mail.freestandards.org/mailman/listinfo/lsb-test