RE: Problems with templates ???

> From: Serge Iovleff [mailto:[email protected]]

> Stefan Seefeld a écrit :
> > Hi Serge,
> > 
> > Here are two little issues with your makefile, though
> > I doubt they cause the error you are seeing:
> > 
> > 
> >># -Wp,-t means  the parser store trailing comments in dummy
> >># -Wp,-b,dir means the parser is using the dir as basename
> > 
> > 
> > there is no '-Wp,-t' nor a '-Wp,-b' option. (As I said, calling
> > conventions changed rather dramatically after 0.5)
> > 
> 
> Sorry, I don't read the doc with care...
> 
> An other bug to solve :
> 
> synopsis  -p Cxx -l JavaComments,JavaTags,SSComments,Previous 
> -Wp,-I../include/ -Wp,--base_path,../include/ -o 
> ../doc/STK_Index.h.syn ../include/STK_Index.h
> Traceback (most recent call last):
>    File "/usr/bin/synopsis", line 181, in ?
>      main()
>    File "/usr/bin/synopsis", line 178, in main
>      processor.process(AST.AST())
>    File 
> "/usr/lib/python2.3/site-packages/Synopsis/Processor.py", 
> line 152, in process
>      ast = self.processors[0].process(ast, **my_kwds)
>    File 
> "/usr/lib/python2.3/site-packages/Synopsis/Parsers/Cxx/Parser.
> py", line 62, in process
>      debug = self.debug)
>    File 
> "/usr/lib/python2.3/site-packages/Synopsis/Parsers/Cpp/Parser.
> py", line 33, in process
>      base_path = self.base_path and 
> os.path.abspath(self.base_path) + os.sep or ''
>    File "/usr/lib/python2.3/posixpath.py", line 395, in abspath
>      if not isabs(path):
>    File "/usr/lib/python2.3/posixpath.py", line 49, in isabs
>      return s.startswith('/')
> AttributeError: 'bool' object has no attribute 'startswith'

ok, that was my fault as I told you the wrong argument syntax.
Calling '-Wp,aaa,bbb,ccc' will result in the parser being
called with arguments 'aaa', 'bbb', and 'ccc'. The argument
you want to pass is '--base_path=dir', and so you should
call '-Wp,--base_path=dir'.
In the above case the parser received the arguments '--base_path' 
and 'dir', i.e. two arguments, not a name/value pair. 
Any argument that doesn't have a value gets set to 'True', 
hence the confusion when python tries to apply the 'startswith' 
method to it (expecting it to be a string).

I admit that this is quite complex, but that's what you end up
with if you want to fine-tune a tool all via command line args
('see, I told you !' ;-)
May be the situation could be enhanced a bit by letting individual
processors check that the arguments have the expected type before
using them...

Regards,
		Stefan