On Thursday, 28 August 2025 at 22:48:16 UTC, Ali Çehreli wrote:
> On 8/28/25 2:51 PM, monkyyy wrote:
> > On Thursday, 28 August 2025 at 21:33:27 UTC, Brother Bill
> wrote:
> >>
> >> The 'old book' pdf seems to be the most current approach,
> and it
> >> should only be about 3 years old.
> >
> > Its not even close to only 3 years old, it mayve been updated
> some but
> > its like 15 years old
>
> That's not an accurate guess: I may have started writing pieces of it 15 years ago but the last update was in February 2022:
>
>   https://bitbucket.org/acehreli/ddili/commits/branch/master
>


Its canonical and probably better then the spec itself, but it is old; updates can only do so much. You got a list of like 50 names your thanking for helping *in 2017*; I think I read it to learn d in 2020, it was quite long then, its been a bit. D itself isnt moving fast so its not like your falling behind, but someone did just go thru your book and found regressions.

> I know I will never get to the level of many programmers

?

Your probably like 5th, your book is canonical given its linked to by official sources enough that I believe every new user is aware of it within a few days.

> but the book was written to teach programming to novices.
> The fact that it uses D is helpful on many levels. There are some old concepts in > there like OOP and strong exception guarantees but I claim the content is still usable today.

My opinions on oop asside, its probably a bit on the heavy side to be a teaching tool, your competing with the spec in my opinion, especially if you only add chapters for new features or clarifying paragraphs as someone disagrees with you. I bet it hurt if someone threw it at ye head.

On Thursday, 28 August 2025 at 22:48:16 UTC, Ali Çehreli wrote:
> That's not an accurate guess: I may have started writing pieces of it 15 years ago but the last update was in February 2022:

While I code in D, a PDF viewer of Mr. Çehreli's book is open in a tab. That the author of such a first-rate resource is available on this forum is... remarkable.

Thank you, Mr. Çehreli!
Andy

On Fri, Aug 29, 2025 at 01:52:59AM +0000, Lance Bachmeier via Digitalmars-d-learn wrote:
> On Thursday, 28 August 2025 at 18:47:19 UTC, Brother Bill wrote:
> > It seems like 'templates' are the 'Achilles heel' of D.
> > 
> > Without starting a flame war, has D gotten to the point where ordinary mortals have difficulty coding in D with 'templates' such as 'cycle' requiring rewrites into 'myCycle'?
> 
> Templates are in general a horrible way to program. Therefore I don't use them very much. When I do, I don't do anything complicated. But everyone has their opinion on that. I similarly don't use attributes. I prefer simple code with a simple language. Hard to avoid ranges in Phobos, which can be crazy complex/inconvenient at times, but every language has its warts.
[...]

Interesting how opinions differ on this. :-D  I couldn't live without templates.  I might be tempted to quit D if I couldn't use templates... ;-) but OTOH there are times when templates are overused where they aren't actually needed.

Once, I wrote an entire program using Phobos ranges, which was amazing for fast prototyping thanks to pipeline programming.  But it resulted in hilariously huge generated symbols (like 4KB for a single symbol!) which slowed compilation down to a crawl while the compiler soaked up RAM like a sponge.  Thankfully, after I complained about it in the forum Rainer implemented a symbol compression scheme, which instantly reduced compilation times back to normal levels.

Then later, while working on a different but somewhat related project, I wanted to reuse the PNG generation modules I'd written entirely in Phobos ranges.  Performance was abysmal, though.  So I ended up rewriting it in more imperative-style code (well OK, with a couple of ranges thrown in here or there 'cos I couldn't be bothered to rewrite every last bit of perfectly fine, non-bottleneck code) and it worked much better.  Thanks to templates and DbI, the transition was smooth and almost without hiccups.  Without them, it would have been a painful exercise in frustration re-fixing bugs that had already been previously fixed.

Mind you, the new code still uses a fair share of templates, but with just the performance-critical bits in more imperative style.  A nice balance between development effort and resulting performance, which is something rarely achieved in other languages.  In a low-level language like C/C++ I'd be pulling my hair out in frustration as I try to navigate language pitfalls and memory micromanagement galore; in a high-level language like Java I'd be cursing the inability to get under the hood and optimize a hot loop without jumping through hoops, needlessly boxed native types, and mountains of boilerplate.  In D? Just a couple o' hours' worth of iteration with a mix of templated and C-like code, and it's already good enough to do real work.  Just a bit more effort and it'd be ready for prime time.  Best of both worlds!


T

-- 
The mathematician suffers from heart disease. He has an arrhythmatic problem.

On Friday, 29 August 2025 at 02:48:21 UTC, H. S. Teoh wrote:
>
> Interesting how opinions differ on this. :-D  I couldn't live without templates.  I might be tempted to quit D if I couldn't use templates... ;-) but OTOH there are times when templates

You probably gonna like comptime in Zig

I think it will be more interesting to compare with more modern languages, than justifying over old ones that are trying to save back-compatibility