Good Lord. Someone else independently came up with something similar to what we came up with a while ago and didn't credit us. How sad. And by the way, they are idiots and we are so much smarter than them. Look, we even wrote a paper.
Indeed, and even worse, lots of people probably used the same algorithm (that is quite basic in the end) for a lot of things before without making a lot of fuss but "we did an academic paper about it, so we consider ourselves as the inventor of it". "How come we don't get credits for being genius".
> Turns out, their concurrency control is a poor-man's version of Left-Right
> Obviously, it's a bit sad when due credit is not given where it should be, but smart readers have added to the blog's comments and on hackernews that what is being described is indeed Left-Right (or a simpler but not as powerful version of it).
> This is ok because the good stuff will be re-discovered several times over, and Left-Right is certainly one of those gold nuggets.
IMHO the tone was quite annoying and overshadowed any good point the article could have about the complexity of proving concurrent algos.
But slightly on a tangent, re: the QuestDB algo, wouldn't hazard pointers solve the concurrent map problem in a cleaner way and avoid a globally contended reader counter?
Good Lord. Someone else independently came up with something similar to what we came up with a while ago and didn't credit us. How sad. And by the way, they are idiots and we are so much smarter than them. Look, we even wrote a paper.
Indeed, and even worse, lots of people probably used the same algorithm (that is quite basic in the end) for a lot of things before without making a lot of fuss but "we did an academic paper about it, so we consider ourselves as the inventor of it". "How come we don't get credits for being genius".
Actually read the article.
That is neither the point the author makes, nor the tone he takes.
> Turns out, their concurrency control is a poor-man's version of Left-Right
> Obviously, it's a bit sad when due credit is not given where it should be, but smart readers have added to the blog's comments and on hackernews that what is being described is indeed Left-Right (or a simpler but not as powerful version of it).
> This is ok because the good stuff will be re-discovered several times over, and Left-Right is certainly one of those gold nuggets.
IMHO the tone was quite annoying and overshadowed any good point the article could have about the complexity of proving concurrent algos.
But slightly on a tangent, re: the QuestDB algo, wouldn't hazard pointers solve the concurrent map problem in a cleaner way and avoid a globally contended reader counter?