The Official Status Thread
-
@topspin said in The Official Status Thread:
@dkf said in The Official Status Thread:
Status: Just linked https://youtu.be/0p_1QSUsbsM in a git commit.
Now I want to see the patch.
Alas, it's a dull one. Renaming constants (admittedly from
MAGIC1
,MAGIC2
, andMAGIC3
) and adding a few explanatory comments. They're all for various command bits in a call that broadcasts an operation (to the various OS nodes) that triggers an application to be terminated.The original names were in Python code written by a colleague, and in turn derived from some arbitrary unnamed values in some even-older Perl code. It's making a little bit of grotty binary encoding just somewhat less awful.
-
@topspin said in The Official Status Thread:
@dkf said in The Official Status Thread:
Status: Just linked https://youtu.be/0p_1QSUsbsM in a git commit.
Now I want to see the patch.
I'd have expected this other Queen song:
https://www.youtube.com/watch?v=UfdAnQ9KXnU
-
@Benjamin-Hall said in The Official Status Thread:
Status Anyone want to suggest materials for doing a deeper dive into SQL wizardry? More about practices for building performant schemas and queries as well as diagnosing bottlenecks than about theory into normalization?
If you're still dealing with MySQL then don't bother, because its optimiser will do whatever the fuck it wants whenever the fuck it wants
@Gribnit said in The Official Status Thread:
First pick a DB. Sadly.
More seriously though this. Plenty of stuff is the same across DB engines but plenty of stuff is different. Especially when it comes to diagnosing bottlenecks/performance issues.
I'd start with something that focuses on one and then worry about the differences when you need to deal with something else.
-
@loopback0 said in The Official Status Thread:
@Benjamin-Hall said in The Official Status Thread:
Status Anyone want to suggest materials for doing a deeper dive into SQL wizardry? More about practices for building performant schemas and queries as well as diagnosing bottlenecks than about theory into normalization?
If you're still dealing with MySQL then don't bother, because its optimiser will do whatever the fuck it wants whenever the fuck it wants
@Gribnit said in The Official Status Thread:
First pick a DB. Sadly.
More seriously though this. Plenty of stuff is the same across DB engines but plenty of stuff is different. Especially when it comes to diagnosing bottlenecks/performance issues.
I'd start with something that focuses on one and then worry about the differences when you need to deal with something else.Currently we're fairly centralized on Aurora in Mysql compatibility mode. But at my current level of knowledge, even the more general stuff would be a good place to start before delving into more specialized stuff.
My "annual goal" is to become the "Database Guy" for the team. And I figure it's a useful skill to have even for my own use (see Lounge for more discussion here).
-
@topspin said in The Official Status Thread:
Status: Snow.
(Damn, that post is late this year!)
Define this year
-
@Applied-Mediocrity this season. I didn’t have time to be pedantically-correct, I needed to post.
-
@HardwareGeek said in The Official Status Thread:
@BernieTheBernie said in The Official Status Thread:
I planned to drive to Frankfurt
But then you'd be in Frankfurt. Bad Ideas thread is .
And driving there is horrible even without snow. Fucking highways with 5 lanes per direction drive (ha) me crazy.
-
@topspin said in The Official Status Thread:
And driving there is horrible even without snow.
I wouldn't know about that. I've only been there twice (if you count an hour or so at the beginning of the trip and an overnight stay before flying out the next morning at the end of the trip separately), and a professional coach driver did all the driving. I just wanted to .
-
@topspin said in The Official Status Thread:
Fucking highways with 5 lanes per direction drive (ha) me crazy.
Oh, you'd just love Silly Valley highways...
-
@Benjamin-Hall said in The Official Status Thread:
@loopback0 said in The Official Status Thread:
@Benjamin-Hall said in The Official Status Thread:
Status Anyone want to suggest materials for doing a deeper dive into SQL wizardry? More about practices for building performant schemas and queries as well as diagnosing bottlenecks than about theory into normalization?
If you're still dealing with MySQL then don't bother, because its optimiser will do whatever the fuck it wants whenever the fuck it wants
@Gribnit said in The Official Status Thread:
First pick a DB. Sadly.
More seriously though this. Plenty of stuff is the same across DB engines but plenty of stuff is different. Especially when it comes to diagnosing bottlenecks/performance issues.
I'd start with something that focuses on one and then worry about the differences when you need to deal with something else.Currently we're fairly centralized on Aurora in Mysql compatibility mode. But at my current level of knowledge, even the more general stuff would be a good place to start before delving into more specialized stuff.
My "annual goal" is to become the "Database Guy" for the team. And I figure it's a useful skill to have even for my own use (see Lounge for more discussion here).
MySQL "compatible" sounds like under the hood is different enough you'd need to find some Aurora-specific material which Amazon may well provide.
-
Status: Bat-Tank at 60 FPS is still disappointing.
-
@Parody Try the new cross-platform PowerShell-tank
https://aka.ms/bscore6
-
@HardwareGeek eh, next thing you do is troll about Berlin and be surprised nobody objects.
-
@topspin said in The Official Status Thread:
@HardwareGeek eh, next thing you do is troll about Berlin and be surprised nobody objects.
I spent 3 days in Berlin, so I'm an expert on that city!
-
status Just moved some rolling drawers so I could move the ethernet switch from one UPS to a different one. The dust bunny back there almost took me out. Wrangled it into the trash can, I need to empty that can now. (Switch was successfully moved)
-
@dcon said in The Official Status Thread:
The dust bunny back there almost took me out. Wrangled it into the trash can, I need to empty that can now.
I am in this post, and I do not like it.
-
@Applied-Mediocrity said in The Official Status Thread:
@topspin You must have been leveraging lots of synergy to forget it like that
I blame this on his assistant มาลัย (meaning "garland of flowers" in Thai)
-
@loopback0 good, good. The sadness increases.
-
@Applied-Mediocrity said in The Official Status Thread:
@topspin said in The Official Status Thread:
Status: Snow.
(Damn, that post is late this year!)
Define this year
The current one. Damn, that was easy. Anybody else have any distinctions being drawn that I have no stake in?
-
@Benjamin-Hall said in The Official Status Thread:
My "annual goal" is to become the "Database Guy" for the team.
Did you... choose... this?
-
@Parody take it down to 4d4 fps, then. Slo-mo will work eventually. You may need to change the audio to something by Queen; which specific song depends of course on the video.
-
@dkf said in The Official Status Thread:
Status: Just linked https://youtu.be/0p_1QSUsbsM in a git commit.
What do you call when you follow a YT link expecting a rick-roll, but it's not?
-
@dcon said in The Official Status Thread:
Just moved some rolling drawers so I could move the ethernet switch from one UPS to a different one. The dust bunny back there almost took me out.
-
@Gribnit said in The Official Status Thread:
@Parody take it down to 4d4 fps, then. Slo-mo will work eventually. You may need to change the audio to something by Queen; which specific song depends of course on the video.
That's roughly what I had the first time around. (Probably 5d4 FPS; I'm pretty sure I hit 20.)
-
@Applied-Mediocrity said in The Official Status Thread:
@Parody Try the new cross-platform PowerShell-tank
https://aka.ms/bscore6Not enough totally non-lethal guns.
-
@Zecc said in The Official Status Thread:
@dkf said in The Official Status Thread:
Status: Just linked https://youtu.be/0p_1QSUsbsM in a git commit.
What do you call when you follow a YT link expecting a rick-roll, but it's not?
A thank-deity moment.
-
@Zerosquare I liked that comic. I was sad when it stopped.
-
status: a BitLocker drive was quick-formatted accidentally. Anyone have good tools to recover it?
-
@Tsaukpaetra what do you need tools for? You are the greatest tool there is.
-
@Gustav said in The Official Status Thread:
@Tsaukpaetra what do you need tools for? You are the greatest tool there is.
I know but I do have limits.
-
-
Status: does cutting your finger on a coffee filter qualify as a workplace accident?
-
I do alright typing and using my touchpad using my non-dominant hand only. But it's still a bit unnerving.
-
@Zecc said in The Official Status Thread:
Status: does cutting your finger on a coffee filter qualify as a workplace accident?
Yes
-
@Zecc said in The Official Status Thread:
does cutting your finger on a coffee filter qualify as a workplace accident?
Only in the workplace, and only if that is not also your home.
-
@dkf said in The Official Status Thread:
@Zecc said in The Official Status Thread:
does cutting your finger on a coffee filter qualify as a workplace accident?
Only in the workplace, and only if that is not also your home.
Actually, no, I think even when working from home it is considered a workplace accident if it is within working hours. Can't be bothered to look into our legislation over a minor cut though.
-
I knew an accident while driving to work was considered a workplace accident, and apparently in Spain that's even true when driving to drop your kids at their grandparent's house before work:
-
Status: Why is this shit so slow?
As part of testing, I have a script that replaces a bunch of lines in a generated file (containing date/time/version etc. making them non-reproducible). The file has an XML-ish format and I basically just run
sed
over it:sed -e "s/Version.*/Version removed/; s/Date.*/Date removed/" -i the_file.dat
I just noticed that for a file with a measly 20MB this takes ~4 minutes to run! What?
Surely, instead of a sed one-liner I can write an extremely naive Python 10-liner doing basically the same, faster... A few minutes later and, of course, without giving any thoughts to efficiency and with Python being generally slow as fuck, this runs in 0.5 seconds!Then I realized the "in place" option (
-i
) is the culprit. So instead:sed -e "s/Version.*/Version removed/; s/Date.*/Date removed/" the_file.dat > the_new_file.dat mv the_new_file.dat the_file.dat
takes 1.5 seconds and does just the same. And that's still 3x slower than the Python version.
-
@topspin said in The Official Status Thread:
Why is this shit so slow?
Have you tried some mild laxative?
-
@topspin said in The Official Status Thread:
Status: Why is this shit so slow?
As part of testing, I have a script that replaces a bunch of lines in a generated file (containing date/time/version etc. making them non-reproducible). The file has an XML-ish format and I basically just run
sed
over it:sed -e "s/Version.*/Version removed/; s/Date.*/Date removed/" -i the_file.dat
I just noticed that for a file with a measly 20MB this takes ~4 minutes to run! What?
Surely, instead of a sed one-liner I can write an extremely naive Python 10-liner doing basically the same, faster... A few minutes later and, of course, without giving any thoughts to efficiency and with Python being generally slow as fuck, this runs in 0.5 seconds!Then I realized the "in place" option (
-i
) is the culprit. So instead:sed -e "s/Version.*/Version removed/; s/Date.*/Date removed/" the_file.dat > the_new_file.dat mv the_new_file.dat the_file.dat
takes 1.5 seconds and does just the same. And that's still 3x slower than the Python version.
Does your file contain newlines?
sed
buffers the entire line by default. Trysed -u
.
-
@PleegWat yes, everything is in relatively short lines (< 60).
sed -u
made it much slower again, as slow as with-i
.
And my naive Python code just did af.readlines()
.
-
@topspin to me, the likely would be use of
;
vs multiple-e
maybe?
-
@topspin What about filtering for lines which need work done without a wildcard, like this. I'm not sure if this applies to sed, but I know
grep
has a separate fixed-string search engine which is way faster than the regex one.sed -e '/Version/s/Version.*/Version removed/' -e '/Date/s/Date.*/Date removed/'
-
@PleegWat no difference, within measurement error. Although even if it did make a difference, it could only explain the 1.5s vs. 0.5s. The interesting part is why writing to a different file and then
mv
ing is so much faster than overwriting a file.
-
@topspin said in The Official Status Thread:
@PleegWat no difference, within measurement error. Although even if it did make a difference, it could only explain the 1.5s vs. 0.5s. The interesting part is why writing to a different file and then
mv
ing is so much faster than overwriting a file.Likely an underpants-on-head implementation of not-a-string-builder that splits the string on either side of the replacement and then splats them together? How many times does the replacement occur? Maybe something terrible with the interoperation of memory-mapped files?
-
Status: I'm sure there's supposed to be more letters and possibly numbers and even symbols here. But I can't be sure...
Windows 11 just hates Chrome that much I guess...
-
And this, folks, is what happens when you mix @Tsaukpaetra and Windows 11. See? Don't do it.
-
@Zerosquare Chrome vs Windows 11 is one of those fights where you hope they both die.
-
@topspin said in The Official Status Thread:
@PleegWat no difference, within measurement error. Although even if it did make a difference, it could only explain the 1.5s vs. 0.5s. The interesting part is why writing to a different file and then
mv
ing is so much faster than overwriting a file.Calling
sync
after every replacement?
Or is the original file on a network drive?
-
@Tsaukpaetra said in The Official Status Thread:
@topspin said in The Official Status Thread:
@PleegWat no difference, within measurement error. Although even if it did make a difference, it could only explain the 1.5s vs. 0.5s. The interesting part is why writing to a different file and then
mv
ing is so much faster than overwriting a file.Likely an underpants-on-head implementation of not-a-string-builder that splits the string on either side of the replacement and then splats them together? How many times does the replacement occur?
It replaces like 3 lines in the first dozen, then the rest just passes through.
@BernieTheBernie said in The Official Status Thread:
@topspin said in The Official Status Thread:
@PleegWat no difference, within measurement error. Although even if it did make a difference, it could only explain the 1.5s vs. 0.5s. The interesting part is why writing to a different file and then
mv
ing is so much faster than overwriting a file.Calling
sync
after every replacement?There really aren't that many ...
Or is the original file on a network drive?
Yes. Who knows what that fucks up.