Constify IDENTIFIER_POINTER; reduce calls to get_identifier

[Gabriel Dos Reis]

Zack Weinberg <zack@wolery.cumb.org> writes:

| On Sat, Aug 26, 2000 at 09:01:37PM +0200, Gabriel Dos Reis wrote:
| > Zack Weinberg <zack@wolery.cumb.org> writes:
| > 
| > [...]
| > 
| > | Bootstrapped i386-linux.  I don't believe there were any regressions
| > | in the test suite, but there's so many "normal" unexpected failures
| > | that it's very hard to tell.  Also, C++ diagnostic reporting is broken
| > | again.
| > 
| > Please, could you expand a little bit on this?  Thanks.
| 
|                 === g++ tests ===
| FAIL: g++.bugs/900210_02.C  (test for errors, line 109)
| FAIL: g++.bugs/900210_02.C (test for excess errors)
| FAIL: g++.ext/instantiate1.C not instantiated (test for errors, line 18)
| FAIL: g++.ext/instantiate1.C not instantiated (test for errors, line 20)
| FAIL: g++.other/loop2.C caused compiler crash
| FAIL: g++.robertl/eb4.C referenced below (test for errors, line 9)
| FAIL: g++.robertl/eb4.C abstract class (test for errors, line 40)
| FAIL: g++.robertl/eb4.C incomplete class (test for errors, line 41)
| FAIL: g++.robertl/eb4.C dfs not declared (test for errors, line 46)
| FAIL: g++.robertl/eb94.C caused compiler crash
| 
| 900210_02.C and eb4.C are the ones I'm talking about.  In 900210_02.C,
| the last error message is truncated:
| 
| /work/src/gcc/gcc/testsuite/g++.old-deja/g++.bugs/900210_02.C:105: cannot convert `signed char' to `void (*) ()' in assignment
| /work/src/gcc/gcc/testsuite/g++.old-deja/g++.bugs/900210_02.C:106: cannot convert `unsigned int' to `void (*) ()' in assignment
| /work/src/gcc/gcc/testsuite/g++.old-deja/g++.bugs/900210_02.C:107: cannot convert `long unsigned int' to `void (*) ()' in assignment
| /work/src/gcc/gcc/testsuite/g++.old-deja/g++.bugs/900210_02.C:108: cannot convert `short unsigned int' to `void (*) ()' in assignment
| /work/src/gcc/gcc/testsuite/g++.old-deja/g++.bugs/
| 
| In eb4.C, all the errors after the first two are missing.  Compare
| last week's tree:

Thanks.  I think this might be caused by system I/O buffering. I just
ran the testsuite  with  

  merlin% g++ -v
  Reading specs from /home/gdr/lib/gcc-lib/i686-pc-linux/2.96/specs
  gcc version 2.96 20000825 (experimental)

but didn't notice the failures you're seeing (apart from
instantiate1.C and loop2.C which are "traditional").  The
regression-tester didn't send a notice either.  I'm building a new
compiler from a freshly updated tree -- I'll let you know the outcome.

-- Gaby
CodeSourcery, LLC                             http://www.codesourcery.com