This is the mail archive of the mailing list for the libstdc++ 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: Getting Apple's libstdc++ debug mode into the FSF tree

Nathan Myers <> writes:

| On Wed, Jul 16, 2003 at 11:02:21PM +0200, Gabriel Dos Reis wrote:
| > Benjamin Kosnik <> writes:
| > 
| > | >I'm in favor of concept checking being on by default, period.
| > | 
| > | You and me both. However, it does impact codegen, or it did the last
| > | time somebody looked closely. I thought that was the reason it was off
| > | by default. (As well as instantiation difficulties on non-ELF platforms,
| > | but perhaps this has been resolved now.)
| > 
| > We also got some reports about false positives.
| That's more troubling than code generation differences.  Still,


| that's a bug report to file.  Again, it's probably already fixed in 
| the codebase.  If concept checking were on by default, the 
| false positives would end up in bugzilla and boost's fix would find 
| its way into libstdc++.
| When I turned it on at my office, our 200KLOC library failed to compile
| because of three errors -- violations of requirements tables -- it 
| detected.  They were easy to fix.  However, I can't get everybody using
| it because on the target machines it's off.  It's off there only because 
| that's the way Red Hat builds it.  They build it that way because that's 
| the default.  (Likewise in Debian.)

Well, there are still packaging problems/education.  System
distributors always have the choice to bundle GCC as they want.
That is no argumennt to say the concept checking should not be on by
default (I'm much worried by the false postives report we got in the
past and instantiation difficulties) but just to point out that
whatever GCC decides, systems packagers have control on build options. 

-- Gaby

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