locale

[Dean Gaudet <dg...@arctic.org>]

Consider PR#76, and PR#679.  I propose that 1.3 does a setlocale(LC_ALL,
"");.  This affects the logging code, but frankly, I don't care.  We can't
be swapping locales around, because it's an expensive call on some
architectures (probably most).  Allow me to curse common log format once
more.  People wanting a workaround can use %m instead of %b in their logs
and teach their log package about numerical month numbers, or post-process
with perl. 

Dean

Re: locale

[Dean Gaudet <dg...@arctic.org>]

Yes well we should document "safe" restarting behaviour anyhow.  Because
the same thing happens now for things like PATH. 

Dean

On Thu, 3 Jul 1997, Marc Slemko wrote:

> On Thu, 3 Jul 1997, Dean Gaudet wrote:
> 
> > It's not that drastic.  For example:
> > 
> > env LANG= httpd -d /serverroot
> 
> Remember, this is users we are dealing with.
> 
> You have someone with their own LANG set restart the server once.  It
> behaves differently.  You reboot.  It behaves differently.  etc...
> 
> > 
> > Dean
> > 
> > On Thu, 3 Jul 1997, Marc Slemko wrote:
> > 
> > > That is a _drastic_ change in behavior for more things than I care to
> > > imagine.  I do not think it is a good thing without a directive or at
> > > least compile time option to enable or disable it.  
> > > 
> > > On Wed, 2 Jul 1997, Dean Gaudet wrote:
> > > 
> > > > Consider PR#76, and PR#679.  I propose that 1.3 does a setlocale(LC_ALL,
> > > > "");.  This affects the logging code, but frankly, I don't care.  We can't
> > > > be swapping locales around, because it's an expensive call on some
> > > > architectures (probably most).  Allow me to curse common log format once
> > > > more.  People wanting a workaround can use %m instead of %b in their logs
> > > > and teach their log package about numerical month numbers, or post-process
> > > > with perl. 
> > > > 
> > > > Dean
> > > > 
> > > > 
> > > 
> > > 
> > 
> 
> 

Re: locale

[Marc Slemko <ma...@worldgate.com>]

On Thu, 3 Jul 1997, Dean Gaudet wrote:

> It's not that drastic.  For example:
> 
> env LANG= httpd -d /serverroot

Remember, this is users we are dealing with.

You have someone with their own LANG set restart the server once.  It
behaves differently.  You reboot.  It behaves differently.  etc...

> 
> Dean
> 
> On Thu, 3 Jul 1997, Marc Slemko wrote:
> 
> > That is a _drastic_ change in behavior for more things than I care to
> > imagine.  I do not think it is a good thing without a directive or at
> > least compile time option to enable or disable it.  
> > 
> > On Wed, 2 Jul 1997, Dean Gaudet wrote:
> > 
> > > Consider PR#76, and PR#679.  I propose that 1.3 does a setlocale(LC_ALL,
> > > "");.  This affects the logging code, but frankly, I don't care.  We can't
> > > be swapping locales around, because it's an expensive call on some
> > > architectures (probably most).  Allow me to curse common log format once
> > > more.  People wanting a workaround can use %m instead of %b in their logs
> > > and teach their log package about numerical month numbers, or post-process
> > > with perl. 
> > > 
> > > Dean
> > > 
> > > 
> > 
> > 
> 

Re: locale

[Dean Gaudet <dg...@arctic.org>]

It's not that drastic.  For example:

env LANG= httpd -d /serverroot

Dean

On Thu, 3 Jul 1997, Marc Slemko wrote:

> That is a _drastic_ change in behavior for more things than I care to
> imagine.  I do not think it is a good thing without a directive or at
> least compile time option to enable or disable it.  
> 
> On Wed, 2 Jul 1997, Dean Gaudet wrote:
> 
> > Consider PR#76, and PR#679.  I propose that 1.3 does a setlocale(LC_ALL,
> > "");.  This affects the logging code, but frankly, I don't care.  We can't
> > be swapping locales around, because it's an expensive call on some
> > architectures (probably most).  Allow me to curse common log format once
> > more.  People wanting a workaround can use %m instead of %b in their logs
> > and teach their log package about numerical month numbers, or post-process
> > with perl. 
> > 
> > Dean
> > 
> > 
> 
> 

Re: locale

[Marc Slemko <ma...@worldgate.com>]

That is a _drastic_ change in behavior for more things than I care to
imagine.  I do not think it is a good thing without a directive or at
least compile time option to enable or disable it.  

On Wed, 2 Jul 1997, Dean Gaudet wrote:

> Consider PR#76, and PR#679.  I propose that 1.3 does a setlocale(LC_ALL,
> "");.  This affects the logging code, but frankly, I don't care.  We can't
> be swapping locales around, because it's an expensive call on some
> architectures (probably most).  Allow me to curse common log format once
> more.  People wanting a workaround can use %m instead of %b in their logs
> and teach their log package about numerical month numbers, or post-process
> with perl. 
> 
> Dean
> 
>