Re: [Axiom-developer] Re: Requiring LaTeX

[Ralf Hemmecke]

On 09/02/2006 12:14 AM, root wrote:
> > Suppose you are an axiom-developer (and you can build everything of
> >  Axiom on your local machine). Now person A is posting a message 
> > that Axiom does not build on machine X for this or that reason. 
> > Let's say the reason is that the binaries don't compile. You have 
> > access to a similar machine X' and you have the feeling that you 
> > could solve the problem.
>
> Here is where the change in MINDSET kicks in.
>
> The source code is [latex vs gcc]
                     ^^^^^^^^^^^^^^
What does that mean?

> and someone posts a bug in a program. You try to build the program 
> and it fails in the [noweb-latex vs gcc compile] step so you give up.
Yes. Because from the description of the build failure posted I had some
feeling what is going wrong. But I cannot reproduce it. Since my build
doesn't come to that point.

> Or you argue that you could take the intermediate code [spad/lisp vs 
> AST-RTL] and debug with that. (GCC can generate intermediate files of
> abstract syntax trees (AST) or register transfer language (RTL) which
> are system-independent.) You can certainly build/debug with that.

I don't even have "intermediate code".

I really don't understand why you insist on preventing people from
contributing to Axiom. It is all community work. If someone understands
what is going on and posts a fix. That is much quicker than having to
wait for someone how has the right environment, understands the problem,
is able to fix it. Note that I could even post a patch that fixes a
problem without having latex on machine X'. Note that I have
(deliberately) included a phrase in parentheses in my first sentence
above. I either do it on my local machine and testing the latex build
there and the binary build on X'. Then I could send a fully documented
patch and everyone would be happy. No?

And even if I submit a patch that does contain a latex error. The guy
who posted the problem could check the binary build and fix the latex
stuff and post a correction patch. That is distributed development! You
need not have just one person knowledgable in everything.

I fully understand that for the enduser you want latex, noweb and
everything, but what does that have to do with developers doing
distributed development to produce proper pamphlets?

> > You checkout Axiom on X' start compilation and... the compilation 
> > stops just after trying to latex the first Makefile.pamphlet, 
> > because unfortunately, LaTeX is not installed on that machine and 
> > you don't have root access to install it.

> So if [noweb-latex vs gcc] is not installed you give up? You don't 
> need root access to install latex. It's just a program. Latex can be 
> installed locally with no problems.

Of course, I could install it. But why should I invest my precious time
with installng LaTeX (or anything else) if it is actually not needed for
what I am trying to achieve.

> The point I'm trying to make is that not have latex is like not 
> having GCC.

Really? If I write a paper, I don't care whether gcc is installed. LaTeX
is what counts. It all depends on what I want to achieve.

> It is perfectly reasonable to have a -nosource option. Many people 
> don't care about the source.
>
> But not having source is just using a binary. It's not a build 
> environment.

There are simply different use cases for Axiom. Some people USE
mathematical theorems but are unable to prove them or even to invent new
ones. What you are saying is: I consider those people not as my target
audience for application of mathematics.

> If build speed is the ONLY issue then just don't 'rm' anything. There
>  is NO reason to rebuild the system from scratch or make clean.

Right. But if I don't have latex and change a pamphlet, I am lost.

> One of the current design principles is that this should 'just work'.

Wasn't there someone who said that --patch-50 does not build on debian?

Ralf