Why can't gcc-4.2.1 build usable libreoffice?

[Mikhail T.]

On 19.02.2013 09:45, Chris Rees wrote:
 >> a. The code is buggy.
 >> b. The compiler is buggy.
 >> c.Both of the above.
 >>My question was, which is it?
 >
 > My answer is that it is almost certainly (b).

Are there identified, known problems with the version? From what little I've 
heard, our cc had some bug-fixes merged-in from newer versions. For example, 
graphics/vigra now compiles fine with the stock cc in 9.1, whereas it used to 
need a newer one.

Maybe, there are already fixes available for whatever is needed for the office 
to build properly as well? The older version may be allowed to miss some 
optimization opportunities or be less descriptive in warnings, but it must 
produce valid binaries from valid code [Captain Obvious hat off]

 > You are welcome to ask upstream about it, but I doubt they would show
 > much interest in such an old compiler.

Upstream gcc? They may not be very interested, indeed, but it is FreeBSD, that 
delivered this compiler to me -- in the most recent stable version of the OS. 
This is why I'm asking stable@'s opinion on the matter...

We aren't really so bad, BTW -- Red Hat Enterprise 5.7 (the latest in their 5.x 
line) still has cc, that identifes itself as:

    gcc version 4.1.2 20080704 (Red Hat 4.1.2-51)

> I think it's insanity that we  still use this version for ports by
 > default, but never mind.

I find it perfectly reasonable, that ports use the base cc and c++ by default. 
But I agree, that it is insane, that the base compiler can not compile one of 
the most popular open-source application-suits...

    -mi