You don't like GC? Do you?
Stanislav Blinov
stanislav.blinov at gmail.com
Fri Oct 12 23:24:56 UTC 2018
On Friday, 12 October 2018 at 21:34:35 UTC, Atila Neves wrote:
>>> -------------------------------
>>> When writing a throwaway script...
>>
>> ...there's absolutely no need for a GC.
>
> True. There's also absolutely no need for computer languages
> either, machine code is sufficient.
Funny. Now for real, in a throwaway script, what is there to gain
from a GC? Allocate away and forget about it.
>> In fact, the GC runtime will only detract from performance.
> Demonstrably untrue. It puzzles me why this myth persists.
Myth, is it now? Unless all you do is allocate memory, which
isn't any kind of useful application, pretty much on each sweep
run the GC's metadata is *cold*. What's worse, you don't control
how much data there is and where it is. Need I say more? If you
disagree, please do the demonstration then.
> There are trade-offs, and one should pick whatever is best for
> the situation at hand.
Exactly. Which is *not at all* what the OP is encouraging to do.
>>> What this means is that whenever I have disregarded a block
>>> of information, say removed an index from an array, then that
>>> memory is automatically cleared and freed back up on the next
>>> sweep. While the process of collection and actually checking
>>
>> Which is just as easily achieved with just one additional line
>> of code: free the memory.
>
> *Simply* achieved, not *easily*. Decades of bugs has shown
> emphatically that it's not easy.
Alright, from one non-native English speaker to another, well
done, I salute you. I also used the term "dangling pointer"
previously, where I should've used "non-null". Strange you didn't
catch that.
To the point: *that* is a myth. The bugs you're referring to are
not *solved* by the GC, they're swept under a rug. Because the
bugs themselves are in the heads, stemming from that proverbial
programmer laziness. It's like everyone is Scarlett O'Hara with a
keyboard.
For most applications, you *do* know how much memory you'll need,
either exactly or an estimation. Garbage collection is useful for
cases when you don't, or can't estimate, and even then a limited
subset of that.
>>> Don't be a computer. Do more with GC.
>>
>> Writing a throwaway script there's nothing stopping you from
>> using mmap or VirtualAlloc.
>
> There is: writing less code to achieve the same result.
Well, I guess either of those do take more arguments than a
"new", so yup, you do indeed write "less" code. Only that you
have no clue how much more code is hiding behind that "new", how
many indirections, DLL calls, syscalls with libc's wonderful
poison that is errno... You don't want to think about that. Then
two people start using your script. Then ten, a hundred, a
thousand. Then it becomes a part of an OS distribution. And no
one wants to "think about that".
>> The "power" of GC is in the language support for non-trivial
>> types, such as strings and associative arrays. Plain old
>> arrays don't benefit from it in the slightest.
> For me, the power of tracing GC is that I don't need to think
> about ownership, lifetimes, or manual memory management.
Yes you do, don't delude yourself. Pretty much the only way you
don't is if you're writing purely functional code. But we're
talking about D here.
Reassigned a reference? You thought about that. If you didn't,
you just wrote a nasty bug. How much more hypocrisy can we reach
here?
"Fun" fact: it's not @safe to "new" anything in D if your program
uses any classes. Thing is, it does unconditionally thanks to
DRuntime.
> I also don't have to please the borrow checker gods.
Yeah, that's another extremum. I guess "rustacians" or whatever
the hell they call themselves are pushing that one, don't they?
"Let's not go for a GC, let's straight up cut out whole paradigms
for safety's sake..."
> Yes, there are other resources to manage. RAII nearly always
> manages that, I don't need to think about that either.
Yes you do. You do need to write those destructors or scoped
finalizers, don't you? Or so help me use a third-party library
that implements those? There's fundamentally *no* difference from
memory management here. None, zero, zip.
Sad thing is, you're not alone. Look at all the major OSs today.
How long does it take to, I don't know, open a project in the
Visual Studio on Windows? Or do a search in a huge file opened in
'less' on Unix? On an octacore 4GHz machine with 32Gb 3GHz
memory? Should just instantly pop up on the screen, shouldn't it?
Why doesn't it then? Because most programmers think the way you
do: "oh it doesn't matter here, I don't need to think about
that". And then proceed to advocate those "awesome" laid-back
solutions that oh so help them save so much time coding. Of
course they do, at everyone else's expense. Decades later, we're
now trying to solve problems that shouldn't have existed in the
first place. You'd think that joke was just that, a joke...
But let's get back to D. Look at Phobos. Why does stdout.writefln
need to allocate? How many times does it copy it's arguments? Why
can't it take non-copyable arguments? Does it change global state
in your process' address space? Does it impose external
dependencies? You don't want to think about that? The author
likely didn't either. And yet everybody is encouraged to use
that: it's out of the box after all...
Why is Socket a class, blown up from a puny 32-bit value to a
bloated who-knows-how-many-bytes monstrosity? Will that socket
close if you rely on the GC? Yes? No? Maybe? Why?
Can I deploy the compiler on a remote machine with limited RAM
and expect it to always successfully build my projects and not
run out of memory?
I can go on and on, but I hope I finally made my point somewhat
clearer. Just in case, a TLDR: *understand your machine and your
tools and use them accordingly*. There are no silver bullets for
anything, and that includes the GC. If you go on advocating it
because it helped you write a 1kLOC one-time-use script, it's
very likely I don't want to use anything you write.
More information about the Digitalmars-d
mailing list