Re: [Axiom-developer] /dev/null, noweb and Axiom

[root]

....[snip]....

> | It's not a question of theology or flavors (or flavours). 
> 
> I'm sorry to say, but a principle followed blinded without taking into
> account reality is theology.
> 
> [...]
> 
> | If you state a principle there is nothing else to explain.
> 
> That *is* theology.

ok. if you insist lets call it theology. i'll be the cleric.
where did i put my stone chisels......



> 
> | The guiding principle in this case is that 
> | 
> |   Axiom never writes outside its own directory
> 
> This *is simply untrue*, at least if you build Axiom+GCL.


it is untrue? i'm confused.

axiom builds gcl in the lsp subdirectory which is contained
within the axiom build tree. 





> 
> [...]
> 
> | > That is close to theology.  /dev/null is very particular and must be
> | > exempted from the general rule.
> | > 
> | 
> | Why?
> 
> Because there simply is no single system on which Axiom has been built
> and where it was impossible to redirect to /dev/null/  That is a
> *fact*.  Your principles must take that fact into account.

and every single system has /tmp but we don't use it.

the ${TMP} directory was created because i ran into a build problem.
another user tried to build Axiom and was using files on /tmp with
the same name. this caused permission problems and caused the build 
to fail. this can't happen anymore.





> 
> | If we assume /dev/null exists we will open ourselves to a porting
> | issue as soon as we attempt a native Windows port.
> 
> Exactly on which windows where will you be able to build Axiom without the
> ability to rediect to /dev/null.  I believe you are trying to solve a
> theoretical problems that has no practical foundation.
> 
> 
> As a data point, there are plenty of software Autoconfiscated that
> build and runs perfectly on windows.


i understand your point that /dev/null seems "ok".
the current ${TMP}/null works.
why change it? did something break?

Father O'Daly.
Chief Cleric and Wearer of Strange Hats.