Please Vote: Exercises in TDPL?

[Denis Koroskin]

On Fri, 15 May 2009 22:09:17 +0400, Andrei Alexandrescu <SeeWebsiteForEmail at erdani.org> wrote:

> Steve Teale wrote:
>> Andrei Alexandrescu Wrote:
>>
>>> A chunky fragment of TDPL will hit Rough Cuts soon enough. I'm  
>>> pondering whether I should be adding exercises to the book. Some books  
>>> have them, some don't.
>>>
>>> Pros: As I'm writing, I've come up with some pretty darn cool exercise  
>>> ideas.
>>>
>>> Cons: The book gets larger, takes longer to write, and I never solved  
>>> the exercises in the books I've read, but then I'm just weird.
>>>
>>> What do you think?
>>>
>>>
>>> Thanks,
>>>
>>> Andrei
>>  Andrei,
>>  Do you have a publisher yet? They will probably give you quite a  
>> definitive answer. They usually have quite strong ideas about the  
>> market they are aiming for.
>
> The publisher, Addison-Wesley, is leaving such details to me.
>
>> If is is to be a university text book, then Yes!
>>  I realize that doing it is a pain in the ass. You have to test every  
>> little thing to the limit, which takes forever. But if you do it you  
>> will sort out bugs in the book.
>>  Don't envy you, especially given the moving target of D2 ;=)
>>  But the best of luck.
>
> Thanks! One nice thing is I've written (in D!) a little script that  
> extracts the code from all of my examples, compiles it, and runs it  
> comparing the output with the expected output. The book will definitely  
> have a number of faults, but code that doesn't work will not be one of  
> them.
>
> It's amazing how much I need to rewrite in wake of recent improvements  
> to D and Phobos. My initial draft of Chapter 1 used char[] for strings!  
> I think D couldn't have claimed being much more than a step forward from  
> C if it stayed with that approach to strings. There's still stuff that  
> doesn't compile (Walter is working on that), and looking forward I'm so  
> excited about the forgotten __traits(allMembers) and the reflection  
> capabilities it begets, I can't stand myself.
>
>
> Andrei

Will you mention __traits as a "__traits" keyword, or it will be renamed (or, better, the whole feature re-implemented) to something better-looking one, by the time TDPL is ready? I really hope it will be fixed in one way or another soon.

We need a better compile-time introspection and I assume __traits is just a starting point. It is too hack-ish and there is definitely a room for improvement.