Re: Use symbol transforms on unlabeled branches?

John Minnihan wrote:
> Summary:
> As a result of a large conversion, I have a number of unlabeled
> branches.   Many other branches depend upon these, and thus these cannot
> be excluded from my conversion.  I'd like to run a symbol transform on
> these to better identify them, using either 
> 
> symbol_transforms=[
>  [...]
>  RegexpSymbolTransform(r'(unlabeled-*)',r'\1/special-name'),
>  [...]

The regexp should probably read r'(unlabeled-.*)', but see below.

> or
> 
> symbol_transforms=[
>  [...]
>  ReplaceSubstringsSymbolTransform('(unlabeled-*)','\1/special-name'),
>  [...]

This wouldn't do anything at all unless you have a symbol containing the
substring "open-paren u n l a b e l e d dash asterisk close-paren".

I don't quite understand the point of doing this.  The name
"unlabeled-*" is already pretty "special" in the sense of easy to spot.
 But more importantly, the names given to unlabeled branches in
different files can overlap even though the branches themselves didn't
have anything to do with each other.  So if by "special-name" you mean
that you want to try to give meaningful names to the unlabeled branches,
then RegexpSymbolTransform and ReplaceSubstringsSymbolTransform are not
powerful enough.  You would need something that can operate at a
file-specific level, like SymbolMapper.  But see below.

> Issue:
> Neither of these produce any visible symbol transformation.  Can
> unlabeled branches be transformed like regular symbols?

That's right; currently, unlabeled branches do not go through the symbol
transformation machinery and therefore cannot be transformed like
regular symbols.  There is no reason that they should not, and it is a
good idea, but it would require some coding to make the change.

Michael

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=1670&dsMessageId=2385800

To unsubscribe from this discussion, e-mail: [[email protected]].