This is the mail archive of the gdb@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Pascal symbols and case sensitivity


> > I think you have to be a little careful of not converting some of the
> > symbols were the "linkage name" is specifically specified. I don't know
> > if this is something that's actually possible with Pascal, but it is
> > something that Ada users sometimes do (Ada is also language insensitive).
> 
> That explains why there is an exception for Ada in symbol_search_name. 

Not quite, actually.  The real issue why Ada needs an exception is that
we do not pre-compute the "natural" names, in order to conserve memory
(we compute the natural names on the fly, and always as temporary memory).
This decision has been at the root of a couple of headaches for me,
particularly performance-wise, but we had no choice, as some of the
projects at the time were too big to fit in the system's memory.

> Besides that I think it would be a good idea to fix the
> case_insensitivity language setting for all languages, so that it's not
> necessary to add it for each language that needs it?

Looks like the problem might be in symtab.c:lookup_symbol_in_language.
Not sure, but the code looks like this:

  if (case_sensitivity == case_sensitive_off)
    {
      char *copy;
      int len, i;

      len = strlen (name);
      copy = (char *) alloca (len + 1);
      for (i= 0; i < len; i++)
        copy[i] = tolower (name[i]);
      copy[len] = 0;
      modified_name = copy;
    }

  returnval = lookup_symbol_aux (modified_name, block, domain, lang,

The case-lowering is done if case_sensitivity is set to off, but this is
the global setting, modified at user level.  It looks like we should be
testing the language's la_case_sensitivity?

> How is it done for Ada? Does Ada use the case_sensitivity language
> setting?

Ada does not set the sensitivity to off - otherwise, we'd be having
issues with symbols where case matters.

Thinking about this so far, I tend to think that a language case-sensitivity
will not be a silver bullet, just because of the few exceptions that might
exist. It's really easy IMO to implement in the language-specific section,
and we could get rid of this language setting which has not been really
used until now.

-- 
Joel


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]