I spent 4 hours building a fucking query for the business team to do some analytics.
I get it working perfectly and one guy jumps on slack, on his phone, assuming that the query doesn’t work and will return duplicates, by UUID.
I tried to explain the query to him. He didn’t understand. I offered to hop on a call and explain it. He wasn’t available.
Finally, I just gpg’d the results over, told him to look at this:
cat Result.csv | cut -d \, -f 1 | uniq -d
I’ve knocked over every conditional and he just says “it’s still not right”
How the fuck is it not right? Because I used subqueries the way he doesn’t like? I returned the right data. I didn’t lock the database. I didn’t knock it over trying to process half a billion records.
Do you ever get that one guy who writes huge sequential queries? Because your database is on ridiculous NVMe storage and it can stream sequential scans at 6 GB/s and they’re all like, “What!? It doesn’t matter, it runs fast like it is.”
Accidentally got a barebones server on ebay that’s incompatible with the E5v2 CPUs I already had to put in it, so now I’m getting E5v3 CPUs too.
Sadly also incompatible with the 80GB of DDR3 reg ECC I ordered to go with the 16GB I already have, so now I’m getting 128GB DDR4 reg ECC and that really hurts.
Anyone need some DDR3 server memory T_T
GitLab 13.0 is going to ship May 22nd and is going to be an exciting release. One of the main features I am excited about is the security scanning like how GitHub does it.
I’ve had luck buying directly from Unix Surplus. If they’re not too busy, they’ll spec something custom for you and they readily have compatibility info on their units.
Well it was my fault for assuming compatibility with v3 implied compat with v2. I did check unixsurplus but didn’t see what I was looking for anyway. Not that I ended up getting what I was looking for. But I’m ending up with something better so I can’t complain too much.
@freqlabs I’ve got a small problem that you might be able to help with, hopefully?
I’m writing out a status line, and writing over it by using \r and just printing the line before. The problem is that when I do that, if the previous line is longer than the next, I get dangling characters. Now, you can just throw a bunch of spaces on the end to clean it up, but that’s a bit hacky. Is there a way to clear the entire line that can be invoked similarly to \r? I’m using python if it helps.
It’s going to depend on the GPU. For Nvidia with the proprietary driver, use nvidia-smi. It can give output in formats which can be parsed by other tools.