50 Bytes of Code That Took 4 GB to Compile (2013)

https://randomascii.wordpress.com/2013/08/14/50-bytes-of-code-that-took-4-gb-to-compile/

by dotnetnews • 9 years ago

102 33 9 years ago

randomascii.wordpress.com

pjc50• 9 years ago

See also previous work in this area: https://news.ycombinator.com/item?id=7127821 "Results of the Grand C++ Error Explosion Competition"

pryelluw2• 9 years ago

@mods

An account with the same username as OP was allegedly spamming r/programming.

https://www.reddit.com/r/programming/comments/5t7akn/50_byte...

DanBC• 9 years ago

You should email the mods (at the address in the guidelines) because they don't see every post and tagging mods doesn't work.

eganist2• 9 years ago

Not even alleged anymore. The account's been shadowbanned (https://www.reddit.com/user/dotnetnews), a confirmation from Reddit admins that they've validated it as a spam/bot account.

And yet this post is back up, which means someone vouched for it.

Flagged it.

endianswap• 9 years ago

I doubt Bruce Dawson commissioned a spam bot to post his ancient articles. My guess is that it was vouched for because it's an actually interesting article (hence the corresponding votes) and users don't really care who posted it if it's interesting.

Sir_Cmpwn2• 9 years ago

Don't judge a post by its author. If the user has done anything to merit being hellbanned on HN, that's a seperate matter. This submission looks fine as-is and there's no reason to flag it on its own merits. I would always vote for telling the user that they're doing something wrong and giving them a chance to correct the behavior before hellbanning them, too.

Users get shadowbanned on Reddit all the time for stupid reasons.

douche• 9 years ago

Seems like a dumb reason to flag something, especially for a new account that's only posted two rather interesting articles here.

garrettgrimsley2• 9 years ago

Is this any different from the educative account [0] that posts nothing but educative.io links? If that account is fine I don't see why this one would not be OK as well. The rules here are not the same as on Reddit. There are no rules specifically about self-promotion.

[0] https://news.ycombinator.com/submitted?id=educative

KirinDave3• 9 years ago

I vouched it, and I am sure I am among others.

1. Reddit's moderators can make their own decisions, but that should not change things here. It's an untrustworthy site with a culture and moderation system that rewards untrustworthy users. If it went out of business tomorrow the world would be a better place.

2. The article is genuinely interesting.

3. If all he bot does is link us to genuinely interesting back catalog .net writing, that's probably a service (at least for now) as this venue has long ignored a lot of interesting content from that ecosystem.

People can break the rules here and then we'll ban them here.

ksk• 9 years ago

Exactly how have you established that it is the same person, and that its a bot?

dang2• 9 years ago

This is a good article. Better still, a good article that was never discussed on HN before. So that account hasn't done anything abusive here—on the contrary.

If abuse occurs, we'll deal with it. If you think abuse might be occurring, please let us know at hn@ycombinator.com. We're can't read all the posts here but we do see all the emails.

pryelluw2• 9 years ago

Thanks. I'll make sure to take the correct steps next time. :)

michaelmrose• 9 years ago

usernames across different websites aren't unique identifiers nor is this relevant to the discussion

Stratoscope2• 9 years ago

...that took 4GB to compile in VS2010.

blackflame70002• 9 years ago

VS2010 is slightly more impactful than other versions because that is the latest version of the IDE before the transition to C++11 standard. The conversion from a 2010 to a 2013 solution is much more complicated than say a 2013 to a 2015 conversion for example.

Stratoscope2• 9 years ago

No disagreement on any of that. I was mostly wondering who uses VS2010 anymore, and curious to know if the bug still exists in newer versions of Visual Studio (but not curious enough to bother testing it).

bluetomcat• 9 years ago

I have accidentally discovered a bug in Clang that causes the compiler generate millions of consecutive "addq $2147483647, %rsp" instructions and it wouldn't terminate maybe in hours (it could probably be an infinite loop, I haven't seen it terminate):

    $ clang --version
    Apple LLVM version 8.0.0 (clang-800.0.38)
    Target: x86_64-apple-darwin16.4.0
    ...
    $ cat test.c
    #include <stdint.h>

    int main(void) {
        int x[SIZE_MAX / 31];
        return 0;
    }
    $ clang test.c
    test.c:5:11: error: array is too large (595056260442243600 elements)
    int x[SIZE_MAX / 31];
    $ cat test-2.c
    #include <stdint.h>

    int main(void) {
        int x[SIZE_MAX / 32];
        return 0;
    }
    $ clang test-2.c
    ... Keep waiting :-) ...

foota• 9 years ago

Strange, seems like it should do that with a loop.

sdardis2• 9 years ago

Interesting, I'm seeing this on clang trunk.

sdardis2• 9 years ago

http://bugs.llvm.org/show_bug.cgi?id=31962

50 Bytes of Code That Took 4 GB to Compile (2013)

Related Stories