Comment on Convicted murderer, filesystem creator writes of regrets to Linux list
maniel@lemmy.ml 10 months ago
Yeah, I was a Linux user at the time, I remember 2.4 to 2.6 kennel transition, it was a breath of fresh air, filesystem wars, some people liked xfs, others jfs, normies used ext3, reiser4 with all the shit they promised, like accessing mp3 on the filesystem level etc was mind-blowing, them he killed his wife…
mozz@mbin.grits.dev 10 months ago
Yeah, around 2.2 - 2.6 time was wild. I was a lurker on LKML at the time and I remember a lot of the same. There was which crazy filesystem to use, there was Alan Cox's huge work like memory or scheduler improvements (I still remember once he started getting it really right I started like 4 compiles in the background and then just went back to working, and it was so responsive still that I forgot about them and left them running), there were whole sagas like ReiserFS or like BitKeeper and the creation of git. Or my all time favorite... CML2.
I remember observing things like Reiser or Bitkeeper play out in real time taught me a lot about how it's not enough to be technically better, you also have to be able to work with people and not be a jerk about things. That's another thing that's great about hearing from Hans, looking back on it all now through the distant lens of hindsight.
maniel@lemmy.ml 10 months ago
Lol, I remember -ac kernel, it has a modified I/O scheduler or something, also I remember using some guys kennel, he went silent one time, after some weeks the guy’s wife wrote on the forum he fell off the ladder, became a vegetable and soon died…
mozz@mbin.grits.dev 10 months ago
Yeah the -ac kernels were for quite some time what the hep cats were all running. Alan Cox did a ton of different performance improvements that slowly made their way into the main kernel over time. I also remember they were way better if you had large amounts of memory for the time.
I also remember this weird little side note when two different teams were both working on some sort of device management subsystem, and when the kernel team selected one and not the other, someone wrote this really touchingly kind note to the other team. Like look, your system is perfectly good, it's easily deserving of getting merged and it's gonna suck that you worked hard on it and it's more or less getting thrown away, but we have to pick and standardize on only one system. But please understand that it's perfectly good and we're not saying it as any kind of value judgement and we hope this doesn't discourage you from contributing good work in the future. It was again that same kind of lesson as with Reiser or BitKeeper that you have to keep the human element in mind.