Incomprehensible compiler errors

[bearophile]

Walter Bright:

Elsewhere you have said:
http://forum.dlang.org/thread/50171A5B.7030002@digitalmars.com

>"DFL won't compile" It doesn't matter if the fix is trivial, 
>easy, whatever. It does not compile, therefore D sux.<

If you have noted the OP has not said D sucks, the opposite:

>The language is fine:<

I agree that having a language perfectly finished is a good 
thing, no one likes broken programs, but programmers can't expect 
D to work as C89, with no changes since many years ago.

A problem is that most people don't understand mangled linker 
errors. I agree with you that linkers are not magic, but modern 
programmers coming from many other languages are not used to them 
and their mangled errors. So we have suggested to filter those 
messages, to help programmers.

The lack of reliable D tools is only partially caused by changes 
in D, another important cause is the small size of the D 
community.

Giving good ways to manage deprecations is a good way to help. 
Giving the programmer ways to fix or solve the problems is 
another way to help.

Recently I have suggested to change the signature of the third 
std.algorithm.count from:
size_t count(alias pred = "true", Range)(Range r);

To:
size_t count(alias pred = "a", Range)(Range r);

But Andrei has refused this tiny change because it breaks old 
code. Introducing ways to manage change & deprecation is going to 
help the development of D more than freezing everything :-)

Bye,
bearophile