Re: DIP 1018--The Copy Constructor--Formal Review
On Thursday, 28 February 2019 at 01:42:13 UTC, Andrei Alexandrescu wrote: Such sharing of resources across objects is a common occurrence, which would be impeded by forcing `const` on the right-hand side of a copy. (An inferior workaround would be to selectively cast `const` away inside the copy constructor, which is obviously undesirable.) For that reason this DIP proposes allowing mutable copy sources. There's an argument to be made that a copy constructor isn't the best way to share resources between two variables in a way that might affect code using the variable being copied.
Re: DIP 1018--The Copy Constructor--Formal Review
On 02/25/2019 03:41 PM, Paolo Invernizzi wrote:
On Monday, 25 February 2019 at 20:23:58 UTC, Andrei Alexandrescu wrote:
On 2/25/19 3:23 PM, Jacob Carlborg wrote:
On 2019-02-25 20:24, Mike Parker wrote:
From the process document:
“the DIP Manager or the Language Maintainers may allow for
exceptions which waive requirements or responsibilities at their
discretion.”
Having it documented doesn't make it less flawed.
Jacob, are there amends you need to make to the DIP?
Honestly, I've not understood the rationale or the covered use case in
letting the copy ctor mutate the ref source parameters...
Sincerely, without polemical intent.
I made an editorial pass:
https://github.com/dlang/DIPs/pull/145
The most material change was additional explanation of using mutable ref
instead of e.g. const. Here's the relevant excerpt, please let me know
if it clarifies the motivation.
=
1. The parameter of the copy constructor is passed by a mutable
reference to the
source object. This means that a call to the copy constructor may
legally modify
the source object:
```d
struct A
{
int[] a;
this(ref A another)
{
another.a[2] = 3;
}
}
void main()
{
A a, b;
a = b;// b.a[2] is modified
}
```
This is surprising and potentially error-prone behavior because changing
the source of a copy is not customary and may surprise the user of a
type. (For that reason, C++ coding standards adopt the convention of
taking the source by means of reference to `const`; copy constructors
that use non-`const` right-hand side are allowed but discouraged.) In D,
`const` and `immutable` are more restrictive than in C++, so forcing
`const` on the copy constructor's right-hand side would make simple
copying task unduly difficult. Consider:
```d
class Window
{
...
}
struct Widget
{
private Window display;
...
this(ref const Widget rhs)
{
display = rhs.display; // Error! Cannot initialize a Window
from a const(Window)
}
}
```
Such sharing of resources across objects is a common occurrence, which
would be impeded by forcing `const` on the right-hand side of a copy.
(An inferior workaround would be to selectively cast `const` away inside
the copy constructor, which is obviously undesirable.) For that reason
this DIP proposes allowing mutable copy sources.
=
Re: DIP 1018--The Copy Constructor--Formal Review
On Tuesday, 26 February 2019 at 05:38:01 UTC, Manu wrote: I'm talking about this DIP. Allowing a mutable copy argument feels super weird. The problem was out of place mutation, which can't happen with copy constructor, because initialization from rvalue is a move, so the copy constructor won't be called.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/27/2019 3:27 AM, Jacob Carlborg wrote: [...] Please start another thread.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2019-02-26 12:07, Walter Bright wrote: It hasn't been approved yet, and please start another thread if you want to discuss it. You have approved the pull request. -- /Jacob Carlborg
Re: DIP 1018--The Copy Constructor--Formal Review
On Tuesday, 26 February 2019 at 03:56:27 UTC, Walter Bright wrote: On 2/25/2019 7:45 AM, Atila Neves wrote: I have no idea what people are talking about when they mention on this forum that D's const is useless. Nearly every function parameter in my code is `in`. Nearly every variable declaration uses `const` instead of `auto`, the main exception being when a function is constructing a value to return. Most member functions I write are const or inout. I use const in D as I do in C++: pretty much everywhere. Nearly every single time a function parameter in my code isn't const is when it's an input range. I'm going to make an educated guess that you're experienced at FP programming! Indeed. Functional is my default in any language. The good thing about D is I can get my hands dirty when needed as opposed to jumping through hoops or forcing me to use recursion when it doesn't fit the problem (hello C++ metaprogramming). I'd love it if functions were `@safe` and `pure` by default.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/26/2019 2:32 AM, Jacob Carlborg wrote: What about the __mutable proposal [1]? [1] https://github.com/dlang/dmd/pull/8315 It hasn't been approved yet, and please start another thread if you want to discuss it.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2019-02-25 01:38, Walter Bright wrote: D's const has teeth. Nothing can be modified through T. What about the __mutable proposal [1]? [1] https://github.com/dlang/dmd/pull/8315 -- /Jacob Carlborg
Re: DIP 1018--The Copy Constructor--Formal Review
On 2019-02-25 21:23, Andrei Alexandrescu wrote: Jacob, are there amends you need to make to the DIP? No, I'm talking about the process in general. -- /Jacob Carlborg
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 02:29:40 UTC, Manu wrote: On Sun, Feb 24, 2019 at 4:25 PM Walter Bright via Digitalmars-d-announce wrote: Thanks for letting me know you're abandoning the rvalue ref DIP. It's not an "rvalue ref" DIP (which I think has confused a lot of people), it's an rvalue *by-ref* DIP. In my head, an "rvalue ref" DIP is something completely different, useful for implementing move semantics of mismatching types. Are you talking about my DIP or that other thing? I had held off working on it because I didn't want to duplicate efforts; we're short-staffed enough as it is. 'abandoning's a strong word, but I don't have motivation to work on it right now. Please, be my guest! The worth of man is not measure by where he stand when all things are going well but where he stands in the time of face of challenge. Your dip has face significant challenge. It is just a test on your ability to forge ahead and get what you believe done in the face of challenges. I will really have problem in taking you serious if you just abandon the dip. Pls rewrite the dip and let it be 50-50% win for all of us
Re: DIP 1018--The Copy Constructor--Formal Review
On Sunday, 24 February 2019 at 10:46:37 UTC, Mike Parker wrote: Walter and Andrei have requested the Final Review round be dropped for DIP 1018, "The Copy Constructor", and have given it their formal approval. They consider copy constructors a critical feature for the language. Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. The DIP: https://github.com/dlang/DIPs/blob/master/DIPs/accepted/DIP1018.md The implementation: https://github.com/dlang/dmd/pull/8688 Are there any other concerns regarding the copy constructor DIP? Although the debate about the value of const's transitivity is an interesting one, it is orthogonal to this proposal. Regarding the mutability of the copy constructor source: it is not mandatory that the source is mutable; you can define it as const, it's just that it is not going to work with mutable objects that contain indirections. There is no workaround for this, and per the current language rules this is not a problem. As for the *by ref* rvalue DIP, I suggest that the discussion should be moved in an other forum thread, as the discussion is hijacked from getting feedback for DIP 1018 Cheers, RazvanN
Re: DIP 1018--The Copy Constructor--Formal Review
On Tuesday, 26 February 2019 at 02:51:54 UTC, Andrei Alexandrescu wrote: On 2/25/19 7:23 PM, Nicholas Wilson wrote: I've said before that that comparison is weak and not particularly useful, irrespective of its intention. That you've said it before does not make it any more correct. You're right that is does't change the facts, thats why it is an opinion. The resemblance, if any, it bears to the processes you describe is completely beside the actual point: namely that it is not _useful_. There are differences, too, of which the public discussions in this forum is the main one. Yes, thats why I think the journal submission metaphor is not apt. This is in danger of getting abused; open discussions around DIPs in this forum give the false impression that DIP authors have the authority to demand any extent of explanation and justification of a decision. Quoting myself from above: On the contrary, if we believe your reasoning to be unsound or misguided (irrespective of who is at fault) then clarification and resolution are the only appropriate courses of action. If you believe that constitutes abuse, I really don't know what to say. We do not have the capacity to do that, and it would not be anymore appropriate than journal reviewers being required to provide detailed feedback to submitters' satisfaction. I sincerely hope you don't lack the capacity, else why are you reviewing it in the first place? Time or bandwidth, sure, thats what DConf Foundation meeting is for (among many other things). This whole notion of a meeting whereby Walter is grilled by a committee on why exactly he rejected DIP 1016 is Kafkaesque. Indeed, that would serve no purpose. What I want from such a meeting is: 1) an outline of requested changes that accurately reflects the sentiments of the DIP. I am fully aware the DIP underspecified and omitted some key aspects and that there were some ambiguities present. You have provided some in that link, thank you. Although I note that some of the advice is outdated/wrong. All this back and forth cements my growing suspicion that the forum is not an appropriate place for such discussions, it severely lacks bandwidth and reciprocal latency. 2) changes to the DIP process, i.e. what to do in the event of DIP breaking behaviour/ambiguities discovered post final review, so that these misunderstandings do not happen again. You can count on me to massage the bureaucracy out of the process if that's the bottleneck. Good, but this one is stuck until DConf, the forum (as a place of communication) lacks the bandwidth for this. The most significant bit is to focus on working together toward making the proposal better, as opposed to focusing on negotiating acceptance. YES! Thats what we've been trying to do! How ever, we can't do that if the criticisms do not reflect reality (lowering to statements, which was given as an example in the DIP and wrongly characterised in the formal assessment) and the ones that are legitimate criticisms are scattered about the forum. Hence why we want the official reasons for rejection at the bottom of the DIP to be updated to reflect that. But whether the DIP keeps the number or gets another one, if the revised document is a 95% cut and paste of the existing one, the review is liable to be a 95% cut and paste of the existing one. You, of all people, should know that a proposal is not a linear function of its text. I _sincerely_ hope that your reviews aren't.
Re: DIP 1018--The Copy Constructor--Formal Review
On Mon, Feb 25, 2019 at 9:30 PM Walter Bright via Digitalmars-d-announce wrote: > > On 2/25/2019 7:17 PM, Manu wrote: > > break my DIP > > The review process is not about "why not add this feature" , but "why should > we > have this feature". > > Additionally, it is most assuredly about finding flaws in it. Isn't it best to > find out the flaws before going further with it than finding them in the > field? > > As I mentioned before, it's supposed to be brutal. Any > testing/certification/review process is about trying to break it. > > It has (hopefully) nothing to do with how hard (or little) you worked on it, > nor > the cut of your jib, nor acceptance (or not) of mediocrity/merit in other > DIPs. I'm talking about this DIP. Allowing a mutable copy argument feels super weird. The reasons are clear, but that doesn't make it feel less weird. I feel like the problem is with const, not with this DIP, but I'm not about to convince anybody, so we're all good here.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 7:17 PM, Manu wrote: break my DIP The review process is not about "why not add this feature" , but "why should we have this feature". Additionally, it is most assuredly about finding flaws in it. Isn't it best to find out the flaws before going further with it than finding them in the field? As I mentioned before, it's supposed to be brutal. Any testing/certification/review process is about trying to break it. It has (hopefully) nothing to do with how hard (or little) you worked on it, nor the cut of your jib, nor acceptance (or not) of mediocrity/merit in other DIPs.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 7:17 PM, Manu wrote: I'm literally astonished that it's been agreed it's fine that a copy constructor can mutate the source... and I can't help but draw contrast to the exact same sorts of arguments that people were using to break my DIP, Mutating the lvalue ref was not the issue, as I recall. It was conversion of the value to a temporary of a different type, then modifying the temporary, not the original. https://digitalmars.com/d/archives/digitalmars/D/announce/DIP_1016--ref_T_accepts_r-values--Formal_Assessment_54145.html#N54345 It is not analogous to the mutable cpctor argument case, because there is no hidden conversion to a temporary.
Re: DIP 1018--The Copy Constructor--Formal Review
On Tuesday, 26 February 2019 at 03:17:53 UTC, Manu wrote: ... It's funny that you keep wasting time posting/replying instead of fixing your DIP. By the way like the other guy, I found disrespectful (With RazvanN7) use this topic as argument for your rejected DIP, instead to debate about what this one has to offer. Donald.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 3:05 PM, Olivier FAURE wrote: On Monday, 25 February 2019 at 22:45:38 UTC, Olivier FAURE wrote: For the same reason C++'s std::shared_pointer uses a non-const copy constructor. Wait, no, I just checked, std::shared_pointer's copy constructor is const, even though it changes shared data. Ugh, that's just wrong. (I kind of agree with Walter's point; I totally assumed the constructor would be non-const, since it mutates data it receives) There's also no requirement that any arguments to constructors should be const. Why should copy-constructors be different?
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 10:44 AM, Andrei Alexandrescu wrote: On 2/25/19 1:39 PM, Jacob Carlborg wrote: On 2019-02-25 17:31, Andrei Alexandrescu wrote: The proposers (Razvan and myself) and Walter (the reviewer) do not know how to make DIP 1018 better. That shouldn't justify accepting a DIP that might contain problems. Definitely. But of course, if you don't think there are any problems. Not talking for Walter, but generally acceptance implies that. I can't guarantee there aren't problems (bugs) in the DIP, but I didn't find any more (the ones I did find Razvan fixed). If you're familiar with the C++ copy ctor, it is very similar, lending credence to the notion that there isn't a big problem left. Regardless, we need to move forward with this, not wring hands wondering what to do.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 6:55 PM, Andrei Alexandrescu wrote: If there's something else I can do to atone, please let me know. Pistols at dawn? Or better yet, why don't I buy both of you a beer at DConf?
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 8:22 AM, H. S. Teoh wrote:
I disagree. Logical const means the outside world can't tell that the
object has changed, because only a constant value is ever seen from
outside. This is the basis of lazy initialization (which is part of the
concept of lazy evaluation), an important feature of FP style code, and
something that D does not support.
We've been through the logical const discussion before. The problem with it is
it is not machine checkable. Therefore it's a documentation thing, not a
language thing.
The not-machine-checkable attributes should be in the domain of user-defined
attributes. It's what they're for.
A derived problem with D's const is the inability to express a cached
const object. You can't cache the object because it's const, which
greatly limits the scope of usability of const in large data structures.
I'd like to draw a distinction between where one can USE const as apposed to
where const has VALUE. I propose that weakening const so it can be used on data
that is mutated anyway means it can be put to USE much more often, at a heavy
price of a severely diminished VALUE. It's why I pressed Manu about the VALUE of
head const.
The same limitation makes ref-counting such a huge challenge to
implement in D. There is simply no way to associate a refcount with a
const object without jumping through hoops and/or treading into UB
territory by casting away const. There is no way to express that the
refcount is mutable but the rest of the object isn't. Well, you *can*
express this if you use circumlocutions like:
struct RefCounted(T) {
int refcount;
const(T) payload;
}
but that's worthless in generic code because const(RefCounted!T) !=
RefCounted!(const T). So you have to special-case every generic function
that needs to work with this type, and the special cases percolate
through the entire codebase, uglifying the code and forcing generic
functions that shouldn't need to know about RefCounted to have to know
about it so that they can work with it.
We're well aware of that issue. The lack of copy-constructors was the first big
barrier to getting ref counting working properly.
Because of these limitations, const is really only useful in low-level
modules of limited scope, in simple, self-contained data structures.
Higher-level, larger data structures are basically unusable with D's
const because lazy initialization and caching are not possible without
treading into UB territory by casting. I'm not going to argue that
C++'s version of const is any better -- because non-enforceable const is
worthless, like you said -- but let's not kid ourselves that D's const
is that much better. D's const is really only usable in very limited
situations, and there are many things for which it's unusable even
though logically it *could* have been applicable.
We'll see. I've been slowly getting better at refactoring code so I can use
const, and as I've recounted before, the results are quite pleasing.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 7:45 AM, Atila Neves wrote: I have no idea what people are talking about when they mention on this forum that D's const is useless. Nearly every function parameter in my code is `in`. Nearly every variable declaration uses `const` instead of `auto`, the main exception being when a function is constructing a value to return. Most member functions I write are const or inout. I use const in D as I do in C++: pretty much everywhere. Nearly every single time a function parameter in my code isn't const is when it's an input range. I'm going to make an educated guess that you're experienced at FP programming!
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 3:09 PM, Olivier FAURE wrote: Yes, this DIP was fast-tracked. Yes, this can feel unfair. And yet, it makes sense that it was fast-tracked, because it fits a priority of the project owners (C++ interoperability + reference counting) and project owners are allowed to have priorities. It's not like this DIP was rushed or has major vulnerabilities (the "mutable copy constructor" thing is necessary for reference counting). And yes, it underwent major rewrites as Razvan can confirm :-)
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/2019 6:26 PM, Andrei Alexandrescu wrote: I haven't heard the final decision from Walter yet, but I proposed that in the interest of quality, we will go through the customary two weeks reviews with DIP 1018. I approved it.
Re: DIP 1018--The Copy Constructor--Formal Review
On Mon, Feb 25, 2019 at 3:10 PM Olivier FAURE via Digitalmars-d-announce wrote: > > On Monday, 25 February 2019 at 16:00:54 UTC, Andrei Alexandrescu > wrote: > > Thorough feedback has been given, likely more so than for any > > other submission. A summary for the recommended steps to take > > can be found here: > > > > https://forum.dlang.org/post/q2u429$1cmg$1...@digitalmars.com > > > > It is not desirable to demand reviewers to do more work on the > > review or to defend it. Acceptance by bullying is unlikely to > > create good results. The target of work is squarely the > > proposal itself. > > Agreed. > > Honestly, I am not impressed with the behavior of several members > here. > > I understand that the rvalue DIP went through a long process, > that some people really wanted it to be accepted, and that it was > frustrating to wait so long only for it to be refused, but at > some point, you guys have to accept that the people in charge > refused it. No, you've missed the point **completely**. I'm not even remotely surprised it was rejected, I never imagined that I'd change peoples minds on this after trying to do so for 10 years running. > They explained why they did, their reasons matched > concerns other users had, and they explained how to move the > proposal forward. This sentence couldn't be more wrong. I'm going to write this again because you prompted me to, I've said it elsewhere lots, but apparently you've missed it; What pissed me off was that the rejection text was almost completely wrong, it almost felt like they just skimmed it and made up details according to presumption, and then when I raised the topic on what was actually wrong looking for actionable feedback, it was made clear that it was not open to amendment, I *must* write a whole new DIP and completely reboot the process because all the text was rubbish, and I should employ someone else competent to do it with me. Then I was insulted a couple more times; it was implied that the DIP was so bad I didn't even understand the implications of my own text (I did), and that it had holes large enough to drive a truck through (it doesn't)... and only then after a few cycles of referring to the *actual* text that was written, it was conceded that those criticisms were indeed incorrect, and then we were able to arrive at some useful feedback, all of which is of trivial-amendment magnitude; fix the rewrite to address exceptions, and add some additional text to clarify a point of misunderstanding that I thought I couldn't have made more clear if I tried. Even at the tail end of that though, the result remained the same: rewrite the DIP, reboot the process, another few hundred days later... it was expressly rejected that an amendment would be accepted for consideration, despite agreeing at the end of the thread that that's all that's required to address the *true* criticisms. That was a worthless experience, and it didn't help anyone. > So again, I get that this is frustrating, but repeatedly > complaining and asking for an appeal and protesting about other > DIPs being accepted is *not* professional behavior. I'm not a bloody professional, I'm a volunteer! I do think it would have been useful to amend the rejection text to be true at the very least, and match the proposal that is written. I held that position before the thread had played out to where useful action points emerged, simply because I wanted to have any idea how to move forward. At the conclusion of that thread, we have the data, and I don't care, although still no path to have it reconsidered with amendments, and I'm not gonna take a few hundred more days to start over. The reason I bring it up here is not that I'm salty (I am), but because I'm literally astonished that it's been agreed it's fine that a copy constructor can mutate the source... and I can't help but draw contrast to the exact same sorts of arguments that people were using to break my DIP, and countless other proposals that I've seen over the years. My DIP was just one of very very many instances of where this class of issue (unexpected mutation of caller-owned data) would be used to destroy something, but we're accepting it here at a very fundamental level of the language. I just can't see how it's fine in this case, after being show-stopping for as long as I've been watching. And to circle right back to the start; I suspect the only reason that it's considered acceptable here, is that this is an issue of extremely high importance, and nobody has any better ideas. To repeat my comment; the problem as I see it, is that `const` as defined is extremely problematic, and rather than address that hard issue, we'll just make a compromise in this case. Anyway, I actually support this DIP, I'm for practical solutions to problems... the only point I was trying to make at the start of this thread is that this sets a precedent, which if we're fair, requires a re-examination of so many rejected ideas gone by.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 9:26 PM, Manu wrote: On Mon, Feb 25, 2019 at 12:20 PM Andrei Alexandrescu via Digitalmars-d-announce wrote: On 2/25/19 2:41 PM, bachmeier wrote: On Monday, 25 February 2019 at 19:24:55 UTC, Mike Parker wrote: From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” If you were to write a DIP for a feature they think important enough, it could be fast tracked, too. I hate to be so negative, but when I see D's corporate management structure, the lack of community contribution is obvious. It doesn't exactly motivate contributions. This is no way to run an open source project. I understand that it works well for Facebook because everyone on the team is paid six figures, and they can be replaced in two hours, but an open source project is not Facebook. I know the whole argument about why it is that way. That doesn't mean it's going to work. What do you recommend? Should we carry a final review here? In my case, you could have produced useful and not-completely-wrong rejection text with the rejection, and then not insulted me a few times before eventually producing some actionable feedback. I mean, its in your interest to foster contribution, not repel it. I apologize again for my use of unkind words. If there's something else I can do to atone, please let me know.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 7:23 PM, Nicholas Wilson wrote: There are similarities and differences between our DIP process and paper submission reviews at conferences and journals everywhere; one key similarity is that the submitters are on hook for providing convincing submissions, whereas reviewers are not required to defend their reviews. It's an asymmetric relationship that occasionally frustrates, but it is as such for good reason and it works. I've said before that that comparison is weak and not particularly useful, irrespective of its intention. That you've said it before does not make it any more correct. We have intently modeled the acceptance process after that used by the review process used by conferences, journals, and standardization committees - naturally from the communities I have some familiarity with (Programming Languages, Machine Learning, Natural Language Processing, Algorithms, ISO C++). So the alleged similarities are more of a statement of fact than a metaphor. There are differences, too, of which the public discussions in this forum is the main one. This is in danger of getting abused; open discussions around DIPs in this forum give the false impression that DIP authors have the authority to demand any extent of explanation and justification of a decision. We do not have the capacity to do that, and it would not be anymore appropriate than journal reviewers being required to provide detailed feedback to submitters' satisfaction. This whole notion of a meeting whereby Walter is grilled by a committee on why exactly he rejected DIP 1016 is Kafkaesque. The keyword here is "short". By suggesting that the action required is to rewrite, the order is most definitely not short. Time is a valuable resource, and a new DIP from scratch through the DIP process takes a lot of it. You can count on me to massage the bureaucracy out of the process if that's the bottleneck. The most significant bit is to focus on working together toward making the proposal better, as opposed to focusing on negotiating acceptance. But whether the DIP keeps the number or gets another one, if the revised document is a 95% cut and paste of the existing one, the review is liable to be a 95% cut and paste of the existing one.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 6:09 PM, Olivier FAURE wrote: Yes, this DIP was fast-tracked. Yes, this can feel unfair. And yet, it makes sense that it was fast-tracked, because it fits a priority of the project owners (C++ interoperability + reference counting) and project owners are allowed to have priorities. It's not like this DIP was rushed or has major vulnerabilities (the "mutable copy constructor" thing is necessary for reference counting). I haven't heard the final decision from Walter yet, but I proposed that in the interest of quality, we will go through the customary two weeks reviews with DIP 1018.
Re: DIP 1018--The Copy Constructor--Formal Review
On Mon, Feb 25, 2019 at 12:20 PM Andrei Alexandrescu via Digitalmars-d-announce wrote: > > On 2/25/19 2:41 PM, bachmeier wrote: > > On Monday, 25 February 2019 at 19:24:55 UTC, Mike Parker wrote: > > > >> From the process document: > >> > >> “the DIP Manager or the Language Maintainers may allow for exceptions > >> which waive requirements or responsibilities at their discretion.” > >> > >> If you were to write a DIP for a feature they think important enough, > >> it could be fast tracked, too. > > > > I hate to be so negative, but when I see D's corporate management > > structure, the lack of community contribution is obvious. It doesn't > > exactly motivate contributions. This is no way to run an open source > > project. I understand that it works well for Facebook because everyone > > on the team is paid six figures, and they can be replaced in two hours, > > but an open source project is not Facebook. > > > > I know the whole argument about why it is that way. That doesn't mean > > it's going to work. > > What do you recommend? Should we carry a final review here? In my case, you could have produced useful and not-completely-wrong rejection text with the rejection, and then not insulted me a few times before eventually producing some actionable feedback. I mean, its in your interest to foster contribution, not repel it.
Re: DIP 1018--The Copy Constructor--Formal Review
On Tuesday, 26 February 2019 at 00:23:19 UTC, Nicholas Wilson wrote: On Monday, 25 February 2019 at 16:00:54 UTC, Andrei Alexandrescu wrote: * etc. etc. etc. That is a good start, though I suspect that the list is not complete given the last item. Oh, it keeps going.
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 20:23:58 UTC, Andrei Alexandrescu wrote: On 2/25/19 3:23 PM, Jacob Carlborg wrote: On 2019-02-25 20:24, Mike Parker wrote: From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” Having it documented doesn't make it less flawed. Jacob, are there amends you need to make to the DIP? This whole reply chain sounds more like the problems lie with the DIP process not this DIP, but I'll let Jacob answer that.
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 16:00:54 UTC, Andrei Alexandrescu wrote: On 2/25/19 1:06 AM, Nicholas Wilson wrote: On Monday, 25 February 2019 at 02:56:13 UTC, Walter Bright wrote: Your DIP, and nobody else is going to do it, so it falls to me. It will be reviewed at Dconf, please make sure you have an _accurate_ summary of your criticisms of the DIP ready for then. This seems to be a misunderstanding of protocol. A negative review is simply a signal that the submission has not been strong enough. As such, the submission, not the review, needs to be improved. I'm not suggesting that the DIP is perfect, nor that it was without ambiguities and misunderstandings, nor that is can't/doesn't need to be improved. What _is_ important is that the time spent on improving it covered that areas that actually need improvement, and given the misunderstandings on all sides, a useful starting point is the set of problems the reviewers have identified crosschecked by the authors. That is not an unreasonable request. There are similarities and differences between our DIP process and paper submission reviews at conferences and journals everywhere; one key similarity is that the submitters are on hook for providing convincing submissions, whereas reviewers are not required to defend their reviews. It's an asymmetric relationship that occasionally frustrates, but it is as such for good reason and it works. I've said before that that comparison is weak and not particularly useful, irrespective of its intention. It is not a matter of misunderstanding 1-2 sentences, but a problem of precision in specification that needs to be approached with due care. I believe it is both, which is the basis for the opinion that resubmission is not the appropriate course of action to make best use of everyone's time. * Description of the typechecking process, with examples of code that passes and code that fails; * A clarification that lowering proceeds not against all expressions, but only against rvalues; * Several places in text in which it is explained that rvalues resulted from implicit conversions are not eligible; * etc. etc. etc. That is a good start, though I suspect that the list is not complete given the last item. It is not desirable to demand reviewers to do more work on the review or to defend it. On the contrary, if we believe your reasoning to be unsound or misguided (irrespective of who is at fault) then clarification and resolution are the only appropriate courses of action. Acceptance by bullying is unlikely to create good results. I agree, but that has not happened here. The target of work is squarely the proposal itself. Our understanding after Manu asked for action items was that he would be up for the work in short order. The keyword here is "short". By suggesting that the action required is to rewrite, the order is most definitely not short. Time is a valuable resource, and a new DIP from scratch through the DIP process takes a lot of it.
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, February 25, 2019 4:09:55 PM MST Olivier FAURE via Digitalmars-d- announce wrote: > Yes, this DIP was fast-tracked. Yes, this can feel unfair. And > yet, it makes sense that it was fast-tracked, because it fits a > priority of the project owners (C++ interoperability + reference > counting) and project owners are allowed to have priorities. It's > not like this DIP was rushed or has major vulnerabilities (the > "mutable copy constructor" thing is necessary for reference > counting). It's worth noting that the copy constructor DIP went through a _lot_ of discussion and was revised accordingly. So, while Walter and Andrei may have considered it a priority, it still took a while for it to get to the point that it was acceptable. - Jonathan M Davis
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 22:45:38 UTC, Olivier FAURE wrote: For the same reason C++'s std::shared_pointer uses a non-const copy constructor. Wait, no, I just checked, std::shared_pointer's copy constructor is const, even though it changes shared data. Ugh, that's just wrong. (I kind of agree with Walter's point; I totally assumed the constructor would be non-const, since it mutates data it receives)
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 16:00:54 UTC, Andrei Alexandrescu wrote: Thorough feedback has been given, likely more so than for any other submission. A summary for the recommended steps to take can be found here: https://forum.dlang.org/post/q2u429$1cmg$1...@digitalmars.com It is not desirable to demand reviewers to do more work on the review or to defend it. Acceptance by bullying is unlikely to create good results. The target of work is squarely the proposal itself. Agreed. Honestly, I am not impressed with the behavior of several members here. I understand that the rvalue DIP went through a long process, that some people really wanted it to be accepted, and that it was frustrating to wait so long only for it to be refused, but at some point, you guys have to accept that the people in charge refused it. They explained why they did, their reasons matched concerns other users had, and they explained how to move the proposal forward. So again, I get that this is frustrating, but repeatedly complaining and asking for an appeal and protesting about other DIPs being accepted is *not* professional behavior. Reviewers are entitled to refuse contributions for any reasons, and if a reviewer rejects a proposal, too bad; you don't get to ask again and again and complain and bring it up in every other thread until they say yes. Yes, this DIP was fast-tracked. Yes, this can feel unfair. And yet, it makes sense that it was fast-tracked, because it fits a priority of the project owners (C++ interoperability + reference counting) and project owners are allowed to have priorities. It's not like this DIP was rushed or has major vulnerabilities (the "mutable copy constructor" thing is necessary for reference counting).
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 20:41:58 UTC, Paolo Invernizzi wrote: Honestly, I've not understood the rationale or the covered use case in letting the copy ctor mutate the ref source parameters... Sincerely, without polemical intent. - P For the same reason C++'s std::shared_pointer uses a non-const copy constructor.
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 20:41:58 UTC, Paolo Invernizzi
wrote:
Honestly, I've not understood the rationale or the covered use
case in letting the copy ctor mutate the ref source
parameters...
Sincerely, without polemical intent.
- P
Because D's const is transitive, you can't copy-construct a
mutable object from a const source:
struct S
{
int[] stuff;
this(const ref S source) {
stuff = source.stuff; // Error: can't assign const(int[])
to int[]
}
}
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 3:41 PM, Paolo Invernizzi wrote: On Monday, 25 February 2019 at 20:23:58 UTC, Andrei Alexandrescu wrote: On 2/25/19 3:23 PM, Jacob Carlborg wrote: On 2019-02-25 20:24, Mike Parker wrote: From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” Having it documented doesn't make it less flawed. Jacob, are there amends you need to make to the DIP? Honestly, I've not understood the rationale or the covered use case in letting the copy ctor mutate the ref source parameters... Sincerely, without polemical intent. We'll look into clarifying that better, thanks.
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 20:23:58 UTC, Andrei Alexandrescu wrote: On 2/25/19 3:23 PM, Jacob Carlborg wrote: On 2019-02-25 20:24, Mike Parker wrote: From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” Having it documented doesn't make it less flawed. Jacob, are there amends you need to make to the DIP? Honestly, I've not understood the rationale or the covered use case in letting the copy ctor mutate the ref source parameters... Sincerely, without polemical intent. - P
Re: DIP 1018--The Copy Constructor--Formal Review
On 2019-02-25 20:24, Mike Parker wrote: From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” Having it documented doesn't make it less flawed. -- /Jacob Carlborg
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 3:23 PM, Jacob Carlborg wrote: On 2019-02-25 20:24, Mike Parker wrote: From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” Having it documented doesn't make it less flawed. Jacob, are there amends you need to make to the DIP?
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 2:41 PM, bachmeier wrote: On Monday, 25 February 2019 at 19:24:55 UTC, Mike Parker wrote: From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” If you were to write a DIP for a feature they think important enough, it could be fast tracked, too. I hate to be so negative, but when I see D's corporate management structure, the lack of community contribution is obvious. It doesn't exactly motivate contributions. This is no way to run an open source project. I understand that it works well for Facebook because everyone on the team is paid six figures, and they can be replaced in two hours, but an open source project is not Facebook. I know the whole argument about why it is that way. That doesn't mean it's going to work. What do you recommend? Should we carry a final review here?
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 19:24:55 UTC, Mike Parker wrote: From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” If you were to write a DIP for a feature they think important enough, it could be fast tracked, too. I hate to be so negative, but when I see D's corporate management structure, the lack of community contribution is obvious. It doesn't exactly motivate contributions. This is no way to run an open source project. I understand that it works well for Facebook because everyone on the team is paid six figures, and they can be replaced in two hours, but an open source project is not Facebook. I know the whole argument about why it is that way. That doesn't mean it's going to work.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 2:24 PM, Mike Parker wrote: On Monday, 25 February 2019 at 18:51:17 UTC, Jacob Carlborg wrote: On 2019-02-24 11:46, Mike Parker wrote: Walter and Andrei have requested the Final Review round be dropped for DIP 1018, "The Copy Constructor", and have given it their formal approval. They consider copy constructors a critical feature for the language. Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. I think the process is flawed when not everyone plays by the same rules. From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” If you were to write a DIP for a feature they think important enough, it could be fast tracked, too. Thanks, Mike. We were a bit surprised ourselves by the scarce response during the community review stage. That said, Jacob and others, if there is a case for the final review please let us know.
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 18:51:17 UTC, Jacob Carlborg wrote: On 2019-02-24 11:46, Mike Parker wrote: Walter and Andrei have requested the Final Review round be dropped for DIP 1018, "The Copy Constructor", and have given it their formal approval. They consider copy constructors a critical feature for the language. Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. I think the process is flawed when not everyone plays by the same rules. From the process document: “the DIP Manager or the Language Maintainers may allow for exceptions which waive requirements or responsibilities at their discretion.” If you were to write a DIP for a feature they think important enough, it could be fast tracked, too.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2019-02-24 11:46, Mike Parker wrote: Walter and Andrei have requested the Final Review round be dropped for DIP 1018, "The Copy Constructor", and have given it their formal approval. They consider copy constructors a critical feature for the language. Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. I think the process is flawed when not everyone plays by the same rules. -- /Jacob Carlborg
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 1:39 PM, Jacob Carlborg wrote: On 2019-02-25 17:31, Andrei Alexandrescu wrote: The proposers (Razvan and myself) and Walter (the reviewer) do not know how to make DIP 1018 better. That shouldn't justify accepting a DIP that might contain problems. Definitely. But of course, if you don't think there are any problems. Not talking for Walter, but generally acceptance implies that.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2019-02-25 17:31, Andrei Alexandrescu wrote: The proposers (Razvan and myself) and Walter (the reviewer) do not know how to make DIP 1018 better. That shouldn't justify accepting a DIP that might contain problems. But of course, if you don't think there are any problems. -- /Jacob Carlborg
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/24/19 4:02 PM, Manu wrote: On Sun, Feb 24, 2019 at 2:50 AM Mike Parker via Digitalmars-d-announce wrote: Walter and Andrei have requested the Final Review round be dropped for DIP 1018, "The Copy Constructor", and have given it their formal approval. They consider copy constructors a critical feature for the language. Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. The DIP: https://github.com/dlang/DIPs/blob/master/DIPs/accepted/DIP1018.md The implementation: https://github.com/dlang/dmd/pull/8688 I mean like, my DIP was almost violently rejected, but in here there's text like this: "The parameter of the copy constructor is passed by a mutable reference to the source object. This means that a call to the copy constructor may legally modify the source object:" I can't see how that could be seen in any way other than what might reasonably be described as "a hole large enough to drive a truck through"... Thanks for the observation. We do not think this creates a hole in the language. Indeed there is risk for misuse but we thought it's better to pass the "true" source of the copy to the constructor instead of affixing additional qualifiers to it. There are important simplicity benefits to doing so because "const" is ambiguous in D as it could originate either from mutable or immutable data. Furthermore, the risk is encapsulated in the type's implementation and not as a misunderstanding between the caller and the callee. I'll add that pointing out a DIP's weaknesses in order to claim an injustice in another DIP is a understandable, but not productive. The proposers (Razvan and myself) and Walter (the reviewer) do not know how to make DIP 1018 better. Walter and I do know, and we have shared, how to make DIP 1016 better. Andrei
Re: DIP 1018--The Copy Constructor--Formal Review
On Sun, Feb 24, 2019 at 08:59:49PM -0800, Walter Bright via
Digitalmars-d-announce wrote:
[...]
> An interesting manifestation of this uselessness in C++ is the notion
> of "logical const", where a supposedly "const" value is lazily set to
> a value upon first use. I.e. it isn't const, it's just pretend const.
I disagree. Logical const means the outside world can't tell that the
object has changed, because only a constant value is ever seen from
outside. This is the basis of lazy initialization (which is part of the
concept of lazy evaluation), an important feature of FP style code, and
something that D does not support.
D's relaxed version of purity -- a function is pure if the outside world
can't see any impure semantics -- makes its scope much more widely
applicable than a strict interpretation of purity as in a functional
language. Logical const is the same idea in the realm of mutability --
though I don't necessarily agree with C++'s anemic implementation of it.
What D could really benefit from was a statically-verifiable way of
lazily initializing something that is const to the outside world.
A derived problem with D's const is the inability to express a cached
const object. You can't cache the object because it's const, which
greatly limits the scope of usability of const in large data structures.
The same limitation makes ref-counting such a huge challenge to
implement in D. There is simply no way to associate a refcount with a
const object without jumping through hoops and/or treading into UB
territory by casting away const. There is no way to express that the
refcount is mutable but the rest of the object isn't. Well, you *can*
express this if you use circumlocutions like:
struct RefCounted(T) {
int refcount;
const(T) payload;
}
but that's worthless in generic code because const(RefCounted!T) !=
RefCounted!(const T). So you have to special-case every generic function
that needs to work with this type, and the special cases percolate
through the entire codebase, uglifying the code and forcing generic
functions that shouldn't need to know about RefCounted to have to know
about it so that they can work with it.
Because of these limitations, const is really only useful in low-level
modules of limited scope, in simple, self-contained data structures.
Higher-level, larger data structures are basically unusable with D's
const because lazy initialization and caching are not possible without
treading into UB territory by casting. I'm not going to argue that
C++'s version of const is any better -- because non-enforceable const is
worthless, like you said -- but let's not kid ourselves that D's const
is that much better. D's const is really only usable in very limited
situations, and there are many things for which it's unusable even
though logically it *could* have been applicable.
T
--
Two wrongs don't make a right; but three rights do make a left...
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 10:45 AM, Atila Neves wrote: The *only* problem I have with const in D is that const values can't be copied, which is silly. I'd expect DIP1018 to fix that. Affirmative. It was tricky.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/25/19 1:06 AM, Nicholas Wilson wrote: On Monday, 25 February 2019 at 02:56:13 UTC, Walter Bright wrote: Your DIP, and nobody else is going to do it, so it falls to me. It will be reviewed at Dconf, please make sure you have an _accurate_ summary of your criticisms of the DIP ready for then. This seems to be a misunderstanding of protocol. A negative review is simply a signal that the submission has not been strong enough. As such, the submission, not the review, needs to be improved. There are similarities and differences between our DIP process and paper submission reviews at conferences and journals everywhere; one key similarity is that the submitters are on hook for providing convincing submissions, whereas reviewers are not required to defend their reviews. It's an asymmetric relationship that occasionally frustrates, but it is as such for good reason and it works. It does happen in such processes that a submission is rejected wrongly, by means of reviews that do not reflect the quality of the submission. This is the case most often when a nonspecialist reviews a specialty paper and fails to appreciate the subtler aspects of the submission. In this DIP's case, we submit this is not the case here; to the extent the DIP failed to convey its intent, that is squarely a pervasive matter with the DIP itself. It is not a matter of misunderstanding 1-2 sentences, but a problem of precision in specification that needs to be approached with due care. Thorough feedback has been given, likely more so than for any other submission. A summary for the recommended steps to take can be found here: https://forum.dlang.org/post/q2u429$1cmg$1...@digitalmars.com It is not desirable to demand reviewers to do more work on the review or to defend it. Acceptance by bullying is unlikely to create good results. The target of work is squarely the proposal itself. Our understanding after Manu asked for action items was that he would be up for the work in short order. Not wanting to step on any toes and seem like taking away credit, Walter decided to hold off on working on it although he wanted to do so now that the matter is in his hands. Now that after three weeks Manu gave permission, I assume Walter would be up for the task. Andrei
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 00:38:02 UTC, Walter Bright wrote: The problem with C++ const is it only goes one level, i.e. what I call "head-const". If you pass a T to a const parameter, anything T references remains mutable. It's more of a suggestion than anything reliable or enforceable. It only works if your data structures are simple aggregates, not graphs. D's const has teeth. Nothing can be modified through T. If you're used to writing code that tweaks const data under the hood, D's const will never work for you. Yes, it means rethinking how the data and code is organized, and that can be painful. But it is how FP works. FP offers a number of advantages, and D's const offers a path into that. I have no idea what people are talking about when they mention on this forum that D's const is useless. Nearly every function parameter in my code is `in`. Nearly every variable declaration uses `const` instead of `auto`, the main exception being when a function is constructing a value to return. Most member functions I write are const or inout. I use const in D as I do in C++: pretty much everywhere. Nearly every single time a function parameter in my code isn't const is when it's an input range. I happened to be working on dpp at the time I wrote this: $ ag 'auto ' source | wc -l 67 $ ag 'const ' source | wc -l 204 The *only* problem I have with const in D is that const values can't be copied, which is silly. I'd expect DIP1018 to fix that.
Re: DIP 1018--The Copy Constructor--Formal Review
On Monday, 25 February 2019 at 02:56:13 UTC, Walter Bright wrote: Your DIP, and nobody else is going to do it, so it falls to me. It will be reviewed at Dconf, please make sure you have an _accurate_ summary of your criticisms of the DIP ready for then. BTW, everyone should expect the DIP process to be brutal. No! It should be thorough, that is a critical distinction. It has to be, we're long past the stage where we can allow misshapen stuff to get into the language. Nobody was very kind with my bottom type DIP :-) Perhaps because you responded to _precisely none_ of the feedback and wasted everybody's time? You should read the dismissal of mine and Andrei's "static if" proposal for C++! (Ironically, it was later adopted after someone else redid it, after removing one of its keystone features.) I have, it is hysterical in all senses to the word. I don't see how that relates to any of this.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/24/2019 8:04 PM, Manu wrote:
1. make code easier to understand?
const code is self documenting and protective against modification by
issuing the user helpful error messages.
I've had many people tell me they mean transitive const when they use const in
C++. That is not self documentation. Nor is it very helpful. The very fact that
const code compiles in C++ and surprises people when it doesn't compile in D
shows the misunderstanding it engenders in C++ code. People infer that it is
transitive when it is not.
Really, of what value is it to know that only the head can't be changed, with no
information about the rest?
An interesting manifestation of this uselessness in C++ is the notion of
"logical const", where a supposedly "const" value is lazily set to a value upon
first use. I.e. it isn't const, it's just pretend const.
2. prevent common programming bugs?
You can't modify const data, for instance,
I know technically what it does. But that wasn't my question.
a copy constructor can't freely modify the source value...
C++ copy constructors are not required to declare the source as const:
"A non-template constructor for class X is a copy constructor if its first
parameter is of type X&, const X&, volatile X& or const volatile X&, and either
there are no other parameters or else all other parameters have default arguments"
-- CPP98.12.8
Again, I know technically what it does. Not the value of it. I used C for a
decade before const was introduced. I tried using it, and discovered that head
const added nothing much of any value and I stopped bothering with it.
3. help with multithreaded coding problems?
This is a different conversation about `immutable` and `shared`.
`const` doesn't say anything about D's decisions relating to
thread-locality by default, which obviously still applies.
'immutable' data is implicitly shareable. Const allows the same functions to
work on both shared immutable and non-shared mutable data. C++ has no way to say
a function can work on both kinds of data, or even how to describe such data.
4. improve code generation?
Not a lot.
Actually, not at all. C++ const is useless for code generation because:
1. people cast away const (yes, they do!). I learned this the hard way and had
to back out those optimizations.
2. other references may mutate it.
Because of (2) it is similarly useless in D, however, it allows the same
functions to work on immutable and mutable data, and that does have value. (The
optimizer can make use of immutable.)
I mean, you speak as if `const` is a synonym for `mutable` in C++...
const things are const.
Not only are they not (legal to cast it away!), they can be mutated by other,
non-const pointers. (There's no way to specify "pointer to immutable" in C++.)
Furthermore, const tells you nothing about the rest of the data structure. C++
const is all fine if your data structure fits entirely in one struct. It's
rather meaningless for more complex objects. You cannot write generic code:
void doSomething(T)(const T x) {...}
and rely on it to not modify T if T is a non-trivial object.
Alternatively, if const were spec-ed similar to C++ const, it would be
very easy to implement TransitiveConst!T as a tool.
You could specify it, but you couldn't use it. (There'd be all kinds of implicit
conversion problems.) There's good reason why const is treated specially for
overload resolution and implicit conversion. I.e.:
void foo(const T const * const * const * p);
const T * const * const * q;
foo(q); // oops!
One thing I quite agree with you on, though, if you really want to modify
non-trivial const objects, then D's const isn't for you :-) If you don't believe
the rewards of carefully designing the data structures and functions that
operate on them so they can be const is worth it, then D's const isn't for you.
For me, I think it is worth it. And yes, it ain't easy, as old habits die hard.
Re: DIP 1018--The Copy Constructor--Formal Review
On Sun, Feb 24, 2019 at 6:35 PM Walter Bright via
Digitalmars-d-announce wrote:
>
> I agree with your point that C++ const can be used in a lot more places than D
> const. Absolutely true.
>
> Missing from the post, however, is an explanation of what value C++ const
> semantics have. How does it:
>
> 1. make code easier to understand?
const code is self documenting and protective against modification by
issuing the user helpful error messages.
> 2. prevent common programming bugs?
You can't modify const data, for instance, a copy constructor can't
freely modify the source value...
> 3. help with multithreaded coding problems?
This is a different conversation about `immutable` and `shared`.
`const` doesn't say anything about D's decisions relating to
thread-locality by default, which obviously still applies.
Maybe you're trying to argue that a const object which contains an
escape-pointer to mutable data may lead to races? But that's not the
case, because all data is thread-local in D, so there's no races on
the mutable data either way unless it's `shared`... and then we need
to refer back to the thread I created months ago where `shared` is
useless and broken, and we REALLY need to fix that. (that is; `shared`
must have NO READ OR WRITE ACCESS to data members, only shared
methods, otherwise it's completely hollow)
> 4. improve code generation?
Not a lot. But this is a red-herring; D's const won't improve code
generation where it's not present in the code.
Contrary to C++, D has a much higher probability of seeing the whole
AST and not encountering opaque extern barriers, which means it would
be relatively easy for D to recognise that the const object contains
no pointers to mutable data (assessed recursively), and then enable
any such optimisations that const offers to D today.
> I know technically what it does (after all, I implemented it), but its value
> escapes me.
I mean, you speak as if `const` is a synonym for `mutable` in C++...
const things are const. It is however possible that they contain a
pointer that leads out of the const data back into the mutable world,
and that's *desirable* in a whole lot of circumstances. Take that
away, and we arrive where we are in D.
It's also easy to NOT have pointers to mutable data escaping const
objects; make them const too!
If you want to implement a semantic where the const-ness of a member
tracks the const-ness of the owner, maybe we can apply `inout` to
behave that way.
Assuming we apply rules similar to C++, it looks like:
const(S) const_s; // const instance of S
struct S
{
int* a; // becomes `int const(*)`
const(int)* b; // const(int*)
inout(int)* c; // becomes const(int*) (or immtable(int*), etc)
}
Alternatively, if const were spec-ed similar to C++ const, it would be
very easy to implement TransitiveConst!T as a tool. By any of these
means, could deploy it deliberately instead of unwillingly.
Re: DIP 1018--The Copy Constructor--Formal Review
Your DIP, and nobody else is going to do it, so it falls to me. BTW, everyone should expect the DIP process to be brutal. It has to be, we're long past the stage where we can allow misshapen stuff to get into the language. Nobody was very kind with my bottom type DIP :-) You should read the dismissal of mine and Andrei's "static if" proposal for C++! (Ironically, it was later adopted after someone else redid it, after removing one of its keystone features.)
Re: DIP 1018--The Copy Constructor--Formal Review
I agree with your point that C++ const can be used in a lot more places than D const. Absolutely true. Missing from the post, however, is an explanation of what value C++ const semantics have. How does it: 1. make code easier to understand? 2. prevent common programming bugs? 3. help with multithreaded coding problems? 4. improve code generation? I know technically what it does (after all, I implemented it), but its value escapes me.
Re: DIP 1018--The Copy Constructor--Formal Review
On Sun, Feb 24, 2019 at 4:25 PM Walter Bright via Digitalmars-d-announce wrote: > > Thanks for letting me know you're abandoning the rvalue ref DIP. It's not an "rvalue ref" DIP (which I think has confused a lot of people), it's an rvalue *by-ref* DIP. In my head, an "rvalue ref" DIP is something completely different, useful for implementing move semantics of mismatching types. Are you talking about my DIP or that other thing? > I had held off > working on it because I didn't want to duplicate efforts; we're short-staffed > enough as it is. 'abandoning's a strong word, but I don't have motivation to work on it right now. Please, be my guest!
Re: DIP 1018--The Copy Constructor--Formal Review
On Sun, Feb 24, 2019 at 4:40 PM Walter Bright via Digitalmars-d-announce wrote: > > The problem with C++ const is it only goes one level, i.e. what I call > "head-const". If you pass a T to a const parameter, anything T references > remains mutable. It's more of a suggestion than anything reliable or > enforceable. It only works if your data structures are simple aggregates, not > graphs. > > D's const has teeth. Nothing can be modified through T. If you're used to > writing code that tweaks const data under the hood, D's const will never work > for you. Yes, it means rethinking how the data and code is organized, and that > can be painful. But it is how FP works. FP offers a number of advantages, and > D's const offers a path into that. > > For example, most of DMD is written in the C++ style where functions > frequently > are written to both return some information *and* tweak the data structure. > This > does not work with const. It needs to be reorganized so that getting > information > about a data structure is separated from modifying the data structure. I've > made > such changes in a few places in DMD, and have been very pleased with the > results > - the code is much easier to understand. > > To sum up, you're quite right that you cannot write C++ style code using D > const. It hast to be in a much more FP style. If you're not accustomed with FP > style, this can be a difficult learning process. I know this from firsthand > experience :-) I agree with these facts, but your case-study is narrow, and you have to stop saying "C++ style", which it really isn't. It's very much D-style... almost all D code is written this way. It's in conflict with too many other patterns, and they're not "C++ patterns", they're very legitimate D patterns. Function pointers and delegates are often incompatible with const; practically any code with some sort of call-back behaviour, and anything that forms *any form* of traversible network where you'd like any part of it to const fails. I've never written a program that was a perfect tree. A small feature library maybe, but not a program that does anything interesting. It's great that we can write FP-ish code in D, it's particularly useful for narrow, self-contained tasks; it helps me intellectually factor some potentially complex leaf-level call-trees out of the program structure, and I appreciate when libraries take that form; it helps them have a smaller footprint in the larger complex suite. But const doesn't play into that much, and if that can't interact with normal D code, which is most code, then it's just not a useful piece of language. The proposition that everyone start writing straight-up FP code in D is unrealistic, and if they wanted that, they'd use Rust every time. People are here because they don't want to write Rust. > For me the only real annoyance with const is I often cannot use "single > assignment" style declarations with pointers: > > I.e.: > > const int* p = > p = // error, good > *p = 4; // also error, not what I wished > > This C++ const does provide, and it's good, but it's not really worth that > much. Are you serious? You can't honestly say C++ const is worthless? Especially in comparison to D's const, which is _actually_ almost completely worthless. It really doesn't make anything better, and there's a whole class of troublesome language issues that emerge from it being defined this way. The way C++ defines const is such that const can be used, and you can integrate that code with other code. I mean it seriously where I say I've tried to defend D's const for as long as I've used D, but I can't escape the plain and honest reality that D's const is not useful for almost anything practical. Even the way you describe it above is like indulging in a little bit of fetish, and I understand that, I try that every time thinking "I'm gonna get it right... this time for sure! What a cool guy I am!", but that never works out beyond a very small scope. const with a narrow scope is where it's least impactful. Then to make matters worse, `const` is a combinatorial testing nightmare; you write your code mostly without const (because conventional wisdom), and then you try and call into your lib from various contexts and it just doesn't work. You need to set-up heaps of tests to try and prove out that your code is const-robust that are very easy to miss otherwise. Then someone else tries to use your code with their code which is using const (attempting at least); I've seen lots of libraries where it would have been possible to support const, at least to some extent, but they just didn't because "don't use const", but the result is that the client of that library can't use const in their own code because the lib undermines their effort in some way. I don't like this concept that a piece of library code 'supports' const, but that's where we are. None of this is issue with C++ const, because it's defined in a way that's useful,
Re: DIP 1018--The Copy Constructor--Formal Review
On 25/02/2019 10:22 AM, Walter Bright wrote: I think we need to throw in the towel, C++'s const is right, and D's const is just wrong, and no amount of pretending that's not true will resolve the endless stream of issues. Where's the DIP to migrate to C++-style const? That is the predicate for basically every important development I've seen lately... including this one. If you want a DIP, write one! But I suggest the rvalue ref one first. We have discussed this on IRC that const needs a redesign. The problem is, without building a list of examples of where it fails we cannot really start the redesign process. But once we do, I'm in.
Re: DIP 1018--The Copy Constructor--Formal Review
On Sunday, 24 February 2019 at 21:22:33 UTC, Walter Bright wrote: On 2/24/2019 1:02 PM, Manu wrote: I mean like, my DIP was almost violently rejected, I thought it was clear what was needed to be done with it, and I thought you were going to rewrite it. Was I mistaken? Absolutely no, no. Yes! "What needs to be done" is scattered throughout forum discussions filled with incorrect assumptions, most of which were resolved needing little change, absolutely nothing that justifies a rewrite over minor amendments (the text will be 95% the same, the sentiment will be 100% the same), neither I nor Manu plan to rewrite it. Note that none of that is reflected in the official review, most (if not all) of which is wrong (and you should update it!). What is _actually_ going to happen is this will be an item on the agenda for the Foundation meeting at dconf: you will clarify the problems you have with the DIP; those problems will be discussed, evaluated and acted upon; and the DIP process will be amended to avoid the root cause of those problems.
Re: DIP 1018--The Copy Constructor--Formal Review
The problem with C++ const is it only goes one level, i.e. what I call "head-const". If you pass a T to a const parameter, anything T references remains mutable. It's more of a suggestion than anything reliable or enforceable. It only works if your data structures are simple aggregates, not graphs. D's const has teeth. Nothing can be modified through T. If you're used to writing code that tweaks const data under the hood, D's const will never work for you. Yes, it means rethinking how the data and code is organized, and that can be painful. But it is how FP works. FP offers a number of advantages, and D's const offers a path into that. For example, most of DMD is written in the C++ style where functions frequently are written to both return some information *and* tweak the data structure. This does not work with const. It needs to be reorganized so that getting information about a data structure is separated from modifying the data structure. I've made such changes in a few places in DMD, and have been very pleased with the results - the code is much easier to understand. To sum up, you're quite right that you cannot write C++ style code using D const. It hast to be in a much more FP style. If you're not accustomed with FP style, this can be a difficult learning process. I know this from firsthand experience :-) For me the only real annoyance with const is I often cannot use "single assignment" style declarations with pointers: I.e.: const int* p = p = // error, good *p = 4; // also error, not what I wished This C++ const does provide, and it's good, but it's not really worth that much.
Re: DIP 1018--The Copy Constructor--Formal Review
Thanks for letting me know you're abandoning the rvalue ref DIP. I had held off working on it because I didn't want to duplicate efforts; we're short-staffed enough as it is.
Re: DIP 1018--The Copy Constructor--Formal Review
On Sun, Feb 24, 2019 at 1:25 PM Walter Bright via Digitalmars-d-announce wrote: > > On 2/24/2019 1:02 PM, Manu wrote: > > I mean like, my DIP was almost violently rejected, > > I thought it was clear what was needed to be done with it, To be fair, initial criticism was 75% just plain wrong (like the text wasn't even read properly, with no request for clarifications), and 100% unproductive. True actionable criticisms became clear only after quite a laborious and somewhat insulting series of exchanges. > and I thought you were going to rewrite it. Was I mistaken? It's not on my short list. I don't really even wanna look at it at this point, my motivation couldn't be more depleted. There's no part of me that has any desire to re-engage that process right now. I'd encourage anybody else to take it and run though. It's still my #1 frustration... it's not getting less annoying! Incidentally, the key problems that upset people about my proposal, and probably the reason it wasn't that way from the very start are all predicated on this same `const` issue. > > but in here there's text like this: > > > > "The parameter of the copy constructor is passed by a mutable > > reference to the source object. This means that a call to the copy > > constructor may legally modify the source object:" > > > > I can't see how that could be seen in any way other than what might > > reasonably be described as "a hole large enough to drive a truck > > through"... > > What's the hole? BTW, the D copy-ctor semantics are nearly identical to that > of C++. Mutable copy-from argument is one of the same arguments people made against my DIP, except about 100x worse being a live object owned by someone else that may be undesirably mutated, rather than an expiring rvalue that nobody will ever see again. I'm mostly just amazed that the same bunch of minds that historically take such strong issue with this sort of thing can find that it's okay in this case... I can't imagine a more concerning case of this class of problem being manifest, but in this case, we've judged that it's fine? If this is acceptable now, then I think it's in order that we comb back over decades of other rejected opportunities and revisit them with this precedent. > > But anyway, that's pretty wild. I think there's a clear pattern we've > > been seeing here with practically every lifetime management DIP, and > > also in general for forever, is that D's `const` just fundamentally > > doesn't work. > > I don't see what const has to do with lifetime management. For example, it is > irrelevant to dip25 and dip1000. I say lifetime *management*; I feel copying/moving and friends are an associated part of lifetime management beyond just tracking ownership. Construction/destruction are features of lifetime management in my brain. We've had const problems with copying and constructors forever, including this DIP, and the problems that this DIP exists for to address. > > Couple this with the prevailing wisdom which is to > > recommend that people "don't use const, because you can't write > > programs and use const" > > That is true for writing C++ style code. D const is much more in line with FP > programming style. It's true for writing D style code; most D-style code is not FP code... at best, a few call-trees at the leaves of the application. The overwhelming recommendation I see posted very frequently in the forum is "don't use const", and the nature of all the articles I've read on the topic as the years progress are moving towards a more clearly stated and unashamed position of "don't use const". I understand the narrow use case where it can be applicable to FP style programming, but it comes up quite infrequently as an opportunity, and attempts are often met with a rude awakening at some point that you work far enough into your project that the fantasy of your flawless design start to break while true details of the program structure begin to emerge. Almost every attempt I've made to try and use D's const effectively has failed at some point down the path as I reach some level of complexity where the program structure has relationships that start to look like a graph. It just naturally occurs that data in a const structure may point back to the outer non-const world again, and that's totally *fine* structurally and intellectually, it's just that D can't express it. You basically have 2 options when this inevitably emerges; you sweep your code removing const from a lot of things (which sadly highlights a whole lot of wasted energy in doing so, and in your foolishly trying in the first place), or you make some HeadConst!(T) thing which casts const away, whereby you deploy UB and a quiet prayer that the compiler doesn't do anything bad. I've tried to defend D's const for a very long time, but the reality is that every sufficiently complex program I've written has seen my attempts with const fail at some level of complexity, and the forum repeats the wisdom
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/24/2019 1:02 PM, Manu wrote: I mean like, my DIP was almost violently rejected, If it makes you feel any better, Razvan had to endure major rewrites of both the dip and the implementation.
Re: DIP 1018--The Copy Constructor--Formal Review
On 2/24/2019 1:02 PM, Manu wrote: I mean like, my DIP was almost violently rejected, I thought it was clear what was needed to be done with it, and I thought you were going to rewrite it. Was I mistaken? but in here there's text like this: "The parameter of the copy constructor is passed by a mutable reference to the source object. This means that a call to the copy constructor may legally modify the source object:" I can't see how that could be seen in any way other than what might reasonably be described as "a hole large enough to drive a truck through"... What's the hole? BTW, the D copy-ctor semantics are nearly identical to that of C++. But anyway, that's pretty wild. I think there's a clear pattern we've been seeing here with practically every lifetime management DIP, and also in general for forever, is that D's `const` just fundamentally doesn't work. I don't see what const has to do with lifetime management. For example, it is irrelevant to dip25 and dip1000. Couple this with the prevailing wisdom which is to recommend that people "don't use const, because you can't write programs and use const" That is true for writing C++ style code. D const is much more in line with FP programming style. I think we need to throw in the towel, C++'s const is right, and D's const is just wrong, and no amount of pretending that's not true will resolve the endless stream of issues. Where's the DIP to migrate to C++-style const? That is the predicate for basically every important development I've seen lately... including this one. If you want a DIP, write one! But I suggest the rvalue ref one first.
Re: DIP 1018--The Copy Constructor--Formal Review
On Sun, Feb 24, 2019 at 2:50 AM Mike Parker via Digitalmars-d-announce wrote: > > Walter and Andrei have requested the Final Review round be > dropped for DIP 1018, "The Copy Constructor", and have given it > their formal approval. They consider copy constructors a critical > feature for the language. > > Walter provided feedback on Razvan's implementation. When it > reached a state with which he was satisfied, he gave the green > light for acceptance. > > The DIP: > https://github.com/dlang/DIPs/blob/master/DIPs/accepted/DIP1018.md > > > The implementation: > https://github.com/dlang/dmd/pull/8688 I mean like, my DIP was almost violently rejected, but in here there's text like this: "The parameter of the copy constructor is passed by a mutable reference to the source object. This means that a call to the copy constructor may legally modify the source object:" I can't see how that could be seen in any way other than what might reasonably be described as "a hole large enough to drive a truck through"... But anyway, that's pretty wild. I think there's a clear pattern we've been seeing here with practically every lifetime management DIP, and also in general for forever, is that D's `const` just fundamentally doesn't work. Couple this with the prevailing wisdom which is to recommend that people "don't use const, because you can't write programs and use const" I think we need to throw in the towel, C++'s const is right, and D's const is just wrong, and no amount of pretending that's not true will resolve the endless stream of issues. Where's the DIP to migrate to C++-style const? That is the predicate for basically every important development I've seen lately... including this one.
Re: DIP 1018--The Copy Constructor--Formal Review
On Sunday, 24 February 2019 at 12:57:06 UTC, ag0aep6g wrote: On 24.02.19 11:46, Mike Parker wrote: Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. Sounds like it might be a "worst acceptable proposal" [1] which Andrei says the DIP process is supposed to avoid. [1] https://forum.dlang.org/post/q2ndr8$15gm$1...@digitalmars.com If I'm understanding correctly, Andrei said that about the proposals, while Walter gave feedback on the implementation, which is a little different. But yeah, the proposal was clearly fast-tracked, probably because it's needed for reference counting and better C++ integration.
Re: DIP 1018--The Copy Constructor--Formal Review
On 24.02.19 11:46, Mike Parker wrote: Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. Sounds like it might be a "worst acceptable proposal" [1] which Andrei says the DIP process is supposed to avoid. [1] https://forum.dlang.org/post/q2ndr8$15gm$1...@digitalmars.com
Re: DIP 1018--The Copy Constructor--Formal Review
On Sunday, 24 February 2019 at 10:46:37 UTC, Mike Parker wrote: Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. The DIP: https://github.com/dlang/DIPs/blob/master/DIPs/accepted/DIP1018.md Yay! Congrats to Razvan. Was this a roadblock towards reference counting, or am I confused?
DIP 1018--The Copy Constructor--Formal Review
Walter and Andrei have requested the Final Review round be dropped for DIP 1018, "The Copy Constructor", and have given it their formal approval. They consider copy constructors a critical feature for the language. Walter provided feedback on Razvan's implementation. When it reached a state with which he was satisfied, he gave the green light for acceptance. The DIP: https://github.com/dlang/DIPs/blob/master/DIPs/accepted/DIP1018.md The implementation: https://github.com/dlang/dmd/pull/8688
