Thread overview
Imperative template patterns
May 05
monkyyy
May 07
user1234
May 07
Monkyyy
May 07
user1234
May 07
Monkyyy
May 07
monkyyy
May 05

As we all should know, the ast is modified live and the compiler never deallocates. This leads to compiler bugs that give glimpses behind the veil. Such bugs features I think are actually quite valuable when the devs are overly cautious, see them disallowing ct io.

Whats the current knowledge of compiler bugs, and how to use them? Anyone found any that Im unaware of?

May 07

On Monday, 5 May 2025 at 22:52:57 UTC, monkyyy wrote:

>

As we all should know, the ast is modified live and the compiler never deallocates. This leads to compiler bugs that give glimpses behind the veil. Such bugs features I think are actually quite valuable when the devs are overly cautious, see them disallowing ct io.

Whats the current knowledge of compiler bugs, and how to use them? Anyone found any that Im unaware of?

mmmh, do you want to know if people are secretly using undocumented compiler bugs like a secret magic knowledge ? That would not be reliable. Someone else can notice the problem, report it, later it gets fixed, and then your exploit does not work anymore.

Sorry if I'm "off the spot" but in first place your question was not very clear.

May 07

On Wednesday, 7 May 2025 at 13:32:41 UTC, user1234 wrote:

>

On Monday, 5 May 2025 at 22:52:57 UTC, monkyyy wrote:

>

As we all should know, the ast is modified live and the compiler never deallocates. This leads to compiler bugs that give glimpses behind the veil. Such bugs features I think are actually quite valuable when the devs are overly cautious, see them disallowing ct io.

Whats the current knowledge of compiler bugs, and how to use them? Anyone found any that Im unaware of?

mmmh, do you want to know if people are secretly using undocumented compiler bugs like a secret magic knowledge ? That would not be reliable. Someone else can notice the problem, report it, later it gets fixed, and then your exploit does not work anymore.

Sorry if I'm "off the spot" but in first place your question was not very clear.

I think several of the bugs I found will never be fixed, probably can't be except by completely disallowing the syntax or major rewrites of systems(the live editing of the ast, is a feature, probably why d is fast while c++ is slow) and there is an compiler fork.

May 07

On Wednesday, 7 May 2025 at 14:49:21 UTC, Monkyyy wrote:

>

On Wednesday, 7 May 2025 at 13:32:41 UTC, user1234 wrote:

>

On Monday, 5 May 2025 at 22:52:57 UTC, monkyyy wrote:

>

As we all should know, the ast is modified live and the compiler never deallocates. This leads to compiler bugs that give glimpses behind the veil. Such bugs features I think are actually quite valuable when the devs are overly cautious, see them disallowing ct io.

Whats the current knowledge of compiler bugs, and how to use them? Anyone found any that Im unaware of?

mmmh, do you want to know if people are secretly using undocumented compiler bugs like a secret magic knowledge ? That would not be reliable. Someone else can notice the problem, report it, later it gets fixed, and then your exploit does not work anymore.

Sorry if I'm "off the spot" but in first place your question was not very clear.

I think several of the bugs I found will never be fixed, probably can't be except by completely disallowing the syntax or major rewrites of systems(the live editing of the ast, is a feature, probably why d is fast while c++ is slow) and there is an compiler fork.

The "live editing of the ast" works (I assume you talk about mixin) because the front-end decomposes analysis in several phases. There are well observed limits however. Not always specified.

May 07

On Wednesday, 7 May 2025 at 14:58:28 UTC, user1234 wrote

>

The "live editing of the ast" works (I assume you talk about mixin)

I don't mean mixin. This is from snar rants and some bugs, but a compiler textbook would want you to have stages of parsing that are pure and sane. Instead it's imparitive, and people complain about "walterisms".

If you priece the veil and find a compile time side effect, it will act surprisingly sanely, and not undecidably.

For reference I suggest the (T, T data) bug, when this as a template header resolves, as an int, the ast starts acting as (T int data) afterwards.

May 07

On Monday, 5 May 2025 at 22:52:57 UTC, monkyyy wrote:

>

As we all should know, the ast is modified live and the compiler never deallocates. This leads to compiler bugs that give glimpses behind the veil. Such bugs features I think are actually quite valuable when the devs are overly cautious, see them disallowing ct io.

Whats the current knowledge of compiler bugs, and how to use them? Anyone found any that Im unaware of?

New code, unlike my old aliaslist doesn't need compiler flags and I believe its should be a sane big O, should be easier to read:

https://gist.github.com/crazymonkyyy/6600a4bb3937e73715093430f7dfb103