I agree, bit-wise comparison is not the way to go, that’s why no compiler does this.
There’s no need to over-engineer this. Keep it simple.


From: Std-Proposals <std-proposals-bounces_at_[hidden]> On Behalf Of Gašper Ažman via Std-Proposals
Sent: Friday, August 30, 2024 11:51 AM
To: marcinjaczewski86_at_[hidden]
Cc: Gašper Ažman <gasper.azman_at_[hidden]>; std-proposals_at_[hidden]; Tom Honermann <tom_at_[hidden]>
Subject: Re: [std-proposals] Revising #pragma once

If you knew up-front you wouldn't do it :).

This happens though. There are people who generate code - whole include trees - and for plugins they end up looking very very similar or identical. And then as the build engineer my users complain that "their build doesn't work on the cloud build environment" and I have to somehow *find their bug*.

Think about it. How the hell do you diagnose this silent noninclusion?

On Fri, Aug 30, 2024 at 9:30 AM Marcin Jaczewski <marcinjaczewski86_at_[hidden]<mailto:marcinjaczewski86_at_[hidden]>> wrote:
pt., 30 sie 2024 o 01:20 Gašper Ažman via Std-Proposals
<std-proposals_at_[hidden]<mailto:std-proposals_at_[hidden]>> napisał(a):
>
> Darlings,
>
> byte-identical is just plain incorrect. Consider.
>
> [library1/public.hpp]
> #pragma once
> #include "utils.hpp"
>
> [library1/utils.hpp]
> class lib1 {};
>
> library2/public.hpp
> #pragma once
> #include "utils.hpp"
>
> [library2/utils.hpp]
> class lib2 {};
>
> [main.cpp]
> #include "library1/public.hpp"
> #include "library2/public.hpp" # boom, library2/utils.hpp does not get included
>
> same-contents also means same-relative-include-trees. Congratulations.
>

Question is do you know upfront that files are problematic?
Then we could do something like this:
a) we only consider normalized paths
b) use have option to add path mappings that are used in path comparison

Like:

We have folders `libs/fooV1` and `libs/fooV2`. Both folders are symlinks
but by default when compiler load `libs/fooV1/bar.h` and `libs/fooV2/bar.h`
the compiler considers them as separate files even when they are the same file.
Now we add compiler flag `-PI libs/fooV1:libs/fooV2` (similar to `-I`) and now
when compiler load `libs/fooV1/foo.h` he consider it as if he load
`libs/fooV2/bar.h` and thus next loading of `libs/fooV2/bar.h` will be blocked.

And this should be ver dumb process it could be cases when `libs/fooV1/bar.h`
and `libs/fooV2/bar.h` are unrelated but if you make specific maping it will
override even diffrnet files. This could be useful to hacking some broken
includes like:

```
-PI hack/foo/bar.h:libs/fooV2/bar.h
```

and we can in our files do:

```
#include "hack/foo/bar.h"
#include "libs/fooV2/foo.h" // it will ignore `#include "bar.h"`
```

Could mechnism like this work on your build system?

> On Fri, Aug 30, 2024 at 12:15 AM Jeremy Rifkin via Std-Proposals <std-proposals_at_[hidden]<mailto:std-proposals_at_[hidden]>> wrote:
>>
>> > In this very thread there are examples showing why taking only the content into account doesn't work but it was brushed off as "that can be fixed".
>>
>> I'm sorry you feel I have brushed any presented examples off. I have
>> found them all immensely helpful for consideration. It's hard for me
>> to imagine times when you'd want the same include-guarded content
>> included twice, however I found the example of a "main header"
>> compelling. The example of a header that only defines macros you undef
>> is also not impractical.
>>
>> However, there are also compelling examples for filesystem identity.
>> Mainly the multiple mount point issue.
>>
>> I think both can be reasonable, however, I have been trying to
>> understand the most probable failure modes. While I originally
>> proposed a content-based definition, I do think a filesystem-based
>> definition is closer to current semantics and expectations.
>>
>> Jeremy
>>
>> On Thu, Aug 29, 2024 at 5:24 PM Breno Guimarães via Std-Proposals
>> <std-proposals_at_[hidden]<mailto:std-proposals_at_[hidden]>> wrote:
>> >
>> > To add to that, the whole idea is to standardize standard practice. If the first thing you do is to change spec to something else, then you're not standardizing standard practice, you are adding a new feature that inconveniently clashes with an existing one.
>> >
>> > In this very thread there are examples showing why taking only the content into account doesn't work but it was brushed off as "that can be fixed".
>> >
>> > None of this make sense to me.
>> >
>> >
>> > Em qui., 29 de ago. de 2024 18:59, Tiago Freire via Std-Proposals <std-proposals_at_[hidden]<mailto:std-proposals_at_[hidden]>> escreveu:
>> >>
>> >> Again, hashing content... totally unnecessary.
>> >>
>> >> There's no need to identify "same content" which as far as I can see can be defeated by modifications that don't change the interpretation, like spaces, which although not technically a violation of "same content" it clearly defeats the intent.
>> >>
>> >> An include summons a resource, a pragma once bars that resources from bey re-summonable. That's it. File paths should be more than enough.
>> >>
>> >> I'm unconvinced that the "bad cases" are not just a product of bad build architecture, if done properly a compiler should never be presented with multiple alternatives of the same file. And putting such requirements on compilers puts an unnecessary burden on developers to support a scenario that it is that is arguably bad practice.
>> >>
>> >> The argument is "prgma once" is supported everywhere it is good, we should make it official in the standard, effectively no change to a compiler should occur as a consequence.
>> >> If a change needs to occur, then in fact "your version" of what you mean by "pragma once" is actually "not supported" by all the major compilers.
>> >>
>> >> Current compiler support of "pragma once" and it's usage on cross platform projects have a particular way of dealing with dependencies in mind. That workflow works. It's pointless to have this discussion if you don't understand that flow, and you shouldn't tailor the tool to a workflow that doesn't exist to the detriment of all.
>> >>
>> >>
>> >> ________________________________
>> >> From: Std-Proposals <std-proposals-bounces_at_[hidden]<mailto:std-proposals-bounces_at_[hidden]>> on behalf of Jeremy Rifkin via Std-Proposals <std-proposals_at_[hidden]<mailto:std-proposals_at_[hidden]>>
>> >> Sent: Thursday, August 29, 2024 9:56:18 PM
>> >> To: Tom Honermann <tom_at_honermann.net<mailto:tom_at_[hidden]>>
>> >> Cc: Jeremy Rifkin <rifkin.jer_at_[hidden]<mailto:rifkin.jer_at_[hidden]>>; std-proposals_at_[hidden]<mailto:std-proposals_at_[hidden]> <std-proposals_at_[hidden]<mailto:std-proposals_at_[hidden]>>
>> >> Subject: Re: [std-proposals] Revising #pragma once
>> >>
>> >> Performance should be fine if using a content definition. An implementation can do inode/path checks against files it already knows of, as a fast path. The first time a file is #included it’s just a hash+table lookup to decide whether to continue.
>> >>
>> >> Regarding the filesystem definition vs content definition question, while I think a content-based definition is robust I can see there is FUD about it and also an argument about current practice being a filesystem-based definition. It may just be best to approach this as filesystem uniqueness to the implementation’s ability, with a requirement that symbolic links/hard links are handled. This doesn’t cover the case of multiple mount points, but we’ve discussed that that’s impossible with #pragma once without using contents instead.
>> >>
>> >> Jeremy
>> >>
>> >> On Thu, Aug 29, 2024 at 13:06 Tom Honermann <tom_at_[hidden]<mailto:tom_at_[hidden]>> wrote:
>> >>>
>> >>> On 8/28/24 12:32 AM, Jeremy Rifkin via Std-Proposals wrote:
>> >>>
>> >>> Another question is whether the comparison should be post translation
>> >>> phase 1.
>> >>>
>> >>> I gave this some thought while drafting the proposal. I think it comes
>> >>> down to whether the intent is single inclusion of files or single
>> >>> inclusion of contents.
>> >>>
>> >>> Indeed. The proposal currently favors the "same contents" approach and offers the following wording.
>> >>>
>> >>> A preprocessing directive of the form
>> >>> # pragma once new-line
>> >>> shall cause no subsequent #include directives to perform replacement for a file with text contents identical to this file.
>> >>>
>> >>> The wording will have to define what it means for contents to be identical. Options include:
>> >>>
>> >>> The files must be byte-for-byte identical. This makes source file encoding observable (which I would be strongly against).
>> >>> The files must encode the same character sequence post translation phase 1. This makes comparisons potentially expensive.
>> >>>
>> >>> Note that the "same contents" approach obligates an implementation to consider every previously encountered file for every #include directive. An inode based optimization can help to determine if a file was previously encountered based on identity, but it doesn't help to reduce the costs when a file that was not previously seen is encountered.
>> >>>
>> >>>
>> >>> Tom.
>> >>>
>> >>> Jeremy
>> >>>
>> >>> On Tue, Aug 27, 2024 at 3:39 PM Tom Honermann via Std-Proposals
>> >>> <std-proposals_at_[hidden]<mailto:std-proposals_at_[hidden]>> wrote:
>> >>>
>> >>> On 8/27/24 4:10 PM, Thiago Macieira via Std-Proposals wrote:
>> >>>
>> >>> On Tuesday 27 August 2024 12:35:17 GMT-7 Andrey Semashev via Std-Proposals
>> >>> wrote:
>> >>>
>> >>> The fact that gcc took the approach to compare file contents I consider
>> >>> a poor choice, and not an argument to standardize this implementation.
>> >>>
>> >>> Another question is whether a byte comparison of two files of the same size is
>> >>> expensive for compilers.
>> >>>
>> >>> #once ID doesn't need to compare the entire file.
>> >>>
>> >>> Another question is whether the comparison should be post translation
>> >>> phase 1. In other words, whether differently encoded source files that
>> >>> decode to the same sequence of code points are considered the same file
>> >>> (e.g., a Windows-1252 version and a UTF-8 version). Standard C++ does
>> >>> not currently allow source file encoding to be observable but a #pragma
>> >>> once implementation that only compares bytes would make such differences
>> >>> observable.
>> >>>
>> >>> Tom.
>> >>>
>> >>> --
>> >>> Std-Proposals mailing list
>> >>> Std-Proposals_at_[hidden]<mailto:Std-Proposals_at_[hidden]>
>> >>> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>> >>
>> >>
>> >> --
>> >> Std-Proposals mailing list
>> >> Std-Proposals_at_[hidden]<mailto:Std-Proposals_at_[hidden]>
>> >> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>> >
>> > --
>> > Std-Proposals mailing list
>> > Std-Proposals_at_[hidden]<mailto:Std-Proposals_at_[hidden]>
>> > https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>> --
>> Std-Proposals mailing list
>> Std-Proposals_at_[hidden]<mailto:Std-Proposals_at_[hidden]>
>> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>
> --
> Std-Proposals mailing list
> Std-Proposals_at_[hidden]<mailto:Std-Proposals_at_[hidden]>
> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals