Re: [RFA] Fix regexp problem with "operator* etc.

Re: [RFA] Fix regexp problem with "operator* etc.

[Michael Snyder]

I committed this change, without Michael Chastain's worthy suggestion
regarding the size of the alloca.  Will submit that separately.

Michael Elizabeth Chastain wrote:
> 
> cplusfuncs.exp:
>   In my maintainer area.
>   Approved unconditionally.
>   No matter what gdb does, the test script should do the right thing.
> 
> alloca change:
>   Outside my maintainer area.
>   Recommended for rework.
> 
>   I think the size is long enough, but the proof of correctness is
>   too delicate.  There is no space explicitly allocated for the final
>   null, so the alloca is correct only because it occurs in a context
>   where at least one character is being deleted.  This also relies on
>   some character-class tests being the same in this function and
>   in operator_chars.   How about something stupidly correct:
> 
>     alloca (8 + fix + strlen(opname) + 1)
> 
> operator_chars change:
>   Outside my maintainer area.
>   Proofread and tested, works for me.
>   Recommended for approval.
> 
> doc/gdb.texinfo:
>   Outside my maintainer area.
> 
>   Recommend add a sentence or two to "@kindex info functions" about functions
>   like "operator[]" and "operator const char *" and how to get at them with
>   "info func operator\[\]" and "info func operator const char \*".
> 
>   Since there is no way at all to type these names in gdb right now,
>   the code can be fixed before the doc is dixed.
> 
> I did my testing on native i686-pc-linux-gnu with gcc 2.95.3 and gcc 3.0.2.
> 
> Michael C

Re: [RFA] Fix regexp problem with "operator* etc.

[Elena Zannoni]

Michael Snyder writes:
 > 
 > I committed this change, without Michael Chastain's worthy suggestion
 > regarding the size of the alloca.  Will submit that separately.
 > 
 > Michael Elizabeth Chastain wrote:
 > > 
 > > cplusfuncs.exp:
 > >   In my maintainer area.
 > >   Approved unconditionally.
 > >   No matter what gdb does, the test script should do the right thing.
 > > 
 > > alloca change:
 > >   Outside my maintainer area.
 > >   Recommended for rework.
 > > 
 > >   I think the size is long enough, but the proof of correctness is
 > >   too delicate.  There is no space explicitly allocated for the final
 > >   null, so the alloca is correct only because it occurs in a context
 > >   where at least one character is being deleted.  This also relies on
 > >   some character-class tests being the same in this function and
 > >   in operator_chars.   How about something stupidly correct:
 > > 
 > >     alloca (8 + fix + strlen(opname) + 1)

Ah, OK. 

 > > 
 > > operator_chars change:
 > >   Outside my maintainer area.
 > >   Proofread and tested, works for me.
 > >   Recommended for approval.

Whoops, I missed this one. Sorry, I'll go and look at it. Did you
commit it already?

Elena



 > > 
 > > doc/gdb.texinfo:
 > >   Outside my maintainer area.
 > > 
 > >   Recommend add a sentence or two to "@kindex info functions" about functions
 > >   like "operator[]" and "operator const char *" and how to get at them with
 > >   "info func operator\[\]" and "info func operator const char \*".
 > > 
 > >   Since there is no way at all to type these names in gdb right now,
 > >   the code can be fixed before the doc is dixed.
 > > 
 > > I did my testing on native i686-pc-linux-gnu with gcc 2.95.3 and gcc 3.0.2.
 > > 
 > > Michael C

Re: [RFA] Fix regexp problem with "operator* etc.

[Michael Snyder]

Elena Zannoni wrote:
> 
> Michael Snyder writes:
>  >
>  > I committed this change, without Michael Chastain's worthy suggestion
>  > regarding the size of the alloca.  Will submit that separately.
>  >
>  > Michael Elizabeth Chastain wrote:
>  > >
>  > > cplusfuncs.exp:
>  > >   In my maintainer area.
>  > >   Approved unconditionally.
>  > >   No matter what gdb does, the test script should do the right thing.
>  > >
>  > > alloca change:
>  > >   Outside my maintainer area.
>  > >   Recommended for rework.
>  > >
>  > >   I think the size is long enough, but the proof of correctness is
>  > >   too delicate.  There is no space explicitly allocated for the final
>  > >   null, so the alloca is correct only because it occurs in a context
>  > >   where at least one character is being deleted.  This also relies on
>  > >   some character-class tests being the same in this function and
>  > >   in operator_chars.   How about something stupidly correct:
>  > >
>  > >     alloca (8 + fix + strlen(opname) + 1)
> 
> Ah, OK.
> 
>  > >
>  > > operator_chars change:
>  > >   Outside my maintainer area.
>  > >   Proofread and tested, works for me.
>  > >   Recommended for approval.
> 
> Whoops, I missed this one. Sorry, I'll go and look at it. Did you
> commit it already?

Yep, I guess I mistook silence for assent.  Sorry if I overstepped.
I'll back it out if it doesn't look OK to you.

Re: [RFA] Fix regexp problem with "operator* etc.

[Elena Zannoni]

Michael Snyder writes:
 > Elena Zannoni wrote:
 > > 
 > > Michael Snyder writes:
 > >  >
 > >  > I committed this change, without Michael Chastain's worthy suggestion
 > >  > regarding the size of the alloca.  Will submit that separately.
 > >  >
 > >  > Michael Elizabeth Chastain wrote:
 > >  > >
 > >  > > cplusfuncs.exp:
 > >  > >   In my maintainer area.
 > >  > >   Approved unconditionally.
 > >  > >   No matter what gdb does, the test script should do the right thing.
 > >  > >
 > >  > > alloca change:
 > >  > >   Outside my maintainer area.
 > >  > >   Recommended for rework.
 > >  > >
 > >  > >   I think the size is long enough, but the proof of correctness is
 > >  > >   too delicate.  There is no space explicitly allocated for the final
 > >  > >   null, so the alloca is correct only because it occurs in a context
 > >  > >   where at least one character is being deleted.  This also relies on
 > >  > >   some character-class tests being the same in this function and
 > >  > >   in operator_chars.   How about something stupidly correct:
 > >  > >
 > >  > >     alloca (8 + fix + strlen(opname) + 1)
 > > 
 > > Ah, OK.
 > > 
 > >  > >
 > >  > > operator_chars change:
 > >  > >   Outside my maintainer area.
 > >  > >   Proofread and tested, works for me.
 > >  > >   Recommended for approval.
 > > 
 > > Whoops, I missed this one. Sorry, I'll go and look at it. Did you
 > > commit it already?
 > 
 > Yep, I guess I mistook silence for assent.  Sorry if I overstepped.
 > I'll back it out if it doesn't look OK to you.


No, no problem. It's fine.

Elena