>>>>> "stevel" == Stephen Lau <stevel at sun.com> writes:
FW> So, how about changing it like this way: FW> 1. Add one argument noident for function hdrchk(fh, filename=None, FW> lenient=False, noident=False) in Check/HdrChk.py FW> 2. Add one new wrapper script/Hdrchk.py to call function hdrchk(..., FW> ..., ..., noident=True). stevel> That sounds like a reasonable solution; though my only qualm stevel> with it is that hdrchk is nominally only supposed to be checking stevel> header files (or at least that's what its name implies), whereas stevel> keywords are a universal thing. It seems like the syntactic checkers need to be contructed in one of 2 ways: (a) the code that determines the SCM is bound in with the checker, or (b) there's one piece that determines the SCM and a second piece that does the actual check. For the (b) approach, the question is then whether the interface is based on "what's the SCM" or on "what features should the checker deal with". Forrest's proposal uses the "what features should the checker deal with" approach. Other checkers could use a similar approach. mike
