|
project conversions are such a busywork idea, nobody gives a care about a commit after it has been brought into main; just copy the files out of the main branch into a new directory and `git init` imo the people who sincerely believe commit messages have value can keep the old repo and never, ever look at it
|
#
?
Oct 18, 2023 23:54
|
|
|
|
| # ? Apr 28, 2024 08:03 |
|
|
oh drat, some dude named Jake committed "add the UI widget, closes #69420" when they wrote this bug. this changes everything and is critical information, thanks for spending your entire 20s copying this over DHH!!!
|
|
#
?
Oct 18, 2023 23:58
|
|
|
how are you going to bisect with that attitude?
|
|
#
?
Oct 19, 2023 00:03
|
|
|
that's why your commit messages should be meaningful aka contain swearing gently caress you I will not squash my commits you're gonna get the full stream of consciousness
|
|
#
?
Oct 19, 2023 00:05
|
|
|
gently caress you I won't Commit What You Tell Me
|
|
#
?
Oct 19, 2023 00:06
|
|
|
Powerful Two-Hander posted:gently caress you I won't Commit What You Tell Me Rage Against the Upstream
|
|
#
?
Oct 19, 2023 00:21
|
|
|
Armitag3 posted:Rage Against the Upstream
|
|
#
?
Oct 19, 2023 00:22
|
|
|
Corla Plankun posted:project conversions are such a busywork idea, nobody gives a care about a commit after it has been brought into main; just copy the files out of the main branch into a new directory and `git init` imo You want it to recover code that someone has overwritten with an older version because the checkout system is trash. Never seen commit messages be used much at all in VSS
|
|
#
?
Oct 19, 2023 00:33
|
|
|
Powerful Two-Hander posted:I want to say source safe was the reason why one repo at my old job was like 1gb for like 2mb of code and it was because people had checked shitlords of msi builds (???) Into source safe and they'd been treated as blobs and just infinitely copied to every subsequent version when it was moved to SVN and nobody could see the ballooning size until I moved it to git and was like "what the gently caress" no one had a problem until we moved to git isnt really the powerful argument against svn that it might seem at first
|
|
#
?
Oct 19, 2023 00:34
|
|
|
Powerful Two-Hander posted:gently caress you I won't Commit What You Tell Me
|
|
#
?
Oct 19, 2023 00:35
|
|
|
like i JUST changed the title
|
|
#
?
Oct 19, 2023 00:36
|
|
|
rotor posted:like i JUST changed the title just force push a change to the commit, that's what mod powers are for
|
|
#
?
Oct 19, 2023 00:39
|
|
|
I made sure to preserve the SVN history when importing projects to Git so that 1) I could stunt on my predecessor, who didn't know how to do this and 2) git blame my predecessor's code and then tell users, "yeah it's been like that for eleven years"
|
|
#
?
Oct 19, 2023 01:22
|
|
|
CPColin posted:I made sure to preserve the SVN history when importing projects to Git so that 1) I could stunt on my predecessor, who didn't know how to do this and 2) git blame my predecessor's code and then tell users, "yeah it's been like that for eleven years" Very good
|
|
#
?
Oct 19, 2023 01:25
|
|
|
Subjunctive posted:how are you going to bisect with that attitude? gently caress bisects. I got biceps
|
|
#
?
Oct 19, 2023 01:44
|
|
|
cut my diffs into pieces this is my pull request
|
|
#
?
Oct 19, 2023 05:32
|
|
|
Armitag3 posted:Rage Against the Upstream
|
|
#
?
Oct 19, 2023 08:14
|
|
|
Armitag3 posted:Rage Against the Upstream
|
|
#
?
Oct 19, 2023 08:33
|
|
|
CPColin posted:I made sure to preserve the SVN history when importing projects to Git so that 1) I could stunt on my predecessor, who didn't know how to do this and 2) git blame my predecessor's code and then tell users, "yeah it's been like that for eleven years" I did exactly the same thing
|
|
#
?
Oct 19, 2023 08:42
|
|
|
CPColin posted:I made sure to preserve the SVN history when importing projects to Git so that 1) I could stunt on my predecessor, who didn't know how to do this and 2) git blame my predecessor's code and then tell users, "yeah it's been like that for eleven years" This is the way
|
|
#
?
Oct 19, 2023 09:14
|
|
|
dbt looks like such a good product. If they ever decided to target the use case that my current org operates in we'd be absolute toast, it's just a much better way of doing things.
|
|
#
?
Oct 19, 2023 09:38
|
|
|
distortion park posted:dbt looks like such a good product. If they ever decided to target the use case that my current org operates in we'd be absolute toast, it's just a much better way of doing things. the concept is brilliant tbh. get people to pay you for letting you turn their sql into jinja2 templates
|
|
#
?
Oct 19, 2023 10:25
|
|
|
CPColin posted:I made sure to preserve the SVN history when importing projects to Git so that 1) I could stunt on my predecessor, who didn't know how to do this and 2) git blame my predecessor's code and then tell users, "yeah it's been like that for eleven years" Porting SVN to Git is a weird process, it could be better explained. I had to manually fix a few dozen changes and I have no idea if I just dropped tonnes of code. It was made worse because the SVN team used a separate folder for each version and I wanted to collapse into a single tree to check across all versions.
|
|
#
?
Oct 19, 2023 15:02
|
|
|
The_Franz posted:that was a pretty common compiler-agnostic idiom to silence pedantic unused parameter warnings for a long time. iirc, gcc/clang had an attribute for this, but msvc needed the x=x nonsense to shut it up. C++17 and C23 finally standardized [[maybe_unused]] the normal way to do it is just a cast to void
|
|
#
?
Oct 19, 2023 18:00
|
|
|
Plorkyeran posted:the normal way to do it is just a cast to void yeah, that's what actual windows headers do. other fun thing mingw decided to do is add extern to the FORCEINLINE macro which breaks static functions. I no longer try to use mingw.
|
|
#
?
Oct 19, 2023 18:47
|
|
|
Plorkyeran posted:the normal way to do it is just a cast to void all C programs should be cast into the void
|
|
#
?
Oct 19, 2023 19:00
|
|
|
Plorkyeran posted:the normal way to do it is just a cast to void c is close to the metal can you point me to the machine code for "cast to void"??? didn't think so
|
|
#
?
Oct 19, 2023 20:00
|
|
|
DELETE CASCADE posted:all C programs should be cast into the void
|
|
#
?
Oct 19, 2023 22:49
|
|
|
I've taken to instrument all asynchronous APIs with execution times, with the eventual goal of feeding into OpenTelemetry. Thus I have a breakdown of CPU processing time and all external interactions, whether that be DB, cache, queue, and special tasks such as image transcoding, video transcoding, checksum calculation, etc. I feel that most people expect their framework of choice to do this all, but that is not something I have seen. The benefits are obvious: it answers the question of wtf is the thing doing, and thus enables informed deployment scaling and targets development at highest reward areas. It enables comparison of different deployment platforms, alongside regression analysis of version changes, for both internal and external code. The downsides are somewhat obvious too: it's a lot of additional data to constantly shuffle around and manage. Weird things this week: (1) There is barely any support for HDR in images, but a lot of effort for video. AVIF seems to be the main winner here, but none of Microsoft tools or even Firefox support it. There are a lot of hack solutions such as 16-bit PNGs, which Netflix uses, and the cesspit of JPEG2000, JPEG-XL, which I'll just ignore. Just to detect that a file is HDR is an overly burdensome amount of effort. Example HDR files here: https://gregbenzphotography.com/hdr/ (2) Oodles of JS APIs support sending HTTP trailers server side, but almost nothing provides convenience functions for client side, i.e. sending content. The primary use case is with AWS S3 for calculating checksums as uploading then sending the checksum as a trailer. The upload needs to be a chunked encoded stream and the trailer manually added. MrMoo fucked around with this message at 02:41 on Oct 20, 2023 |
|
#
?
Oct 20, 2023 01:16
|
|
|
https://www.youtube.com/watch?v=vxmZPMg7vIs Microsoft Japan and Ruby, Japan's software industry has a lot to answer for. MrMoo fucked around with this message at 04:14 on Oct 20, 2023 |
|
#
?
Oct 20, 2023 04:06
|
|
|
Oh man I'm looking forward to that full interview. Love me some Dave Cutler.
|
|
#
?
Oct 20, 2023 05:37
|
|
|
MrMoo posted:(1) There is barely any support for HDR in images, but a lot of effort for video. AVIF seems to be the main winner here, but none of Microsoft tools or even Firefox support it. There are a lot of hack solutions such as 16-bit PNGs, which Netflix uses, and the cesspit of JPEG2000, JPEG-XL, which I'll just ignore. Just to detect that a file is HDR is an overly burdensome amount of effort. just use EXRs 
|
|
#
?
Oct 20, 2023 08:36
|
|
|
MrMoo posted:I've taken to instrument all asynchronous APIs with execution times, with the eventual goal of feeding into OpenTelemetry. Thus I have a breakdown of CPU processing time and all external interactions, whether that be DB, cache, queue, and special tasks such as image transcoding, video transcoding, checksum calculation, etc. Datadog makes this real easy if you can write them a nice big monthly check
|
|
#
?
Oct 20, 2023 08:55
|
|
|
distortion park posted:Datadog makes this real easy if you can write them a nice big monthly check and however big you think the check will be, it will be slightly bigger but their poo poo really is very nice
|
|
#
?
Oct 20, 2023 09:26
|
|
|
Found this commit title today:quote:Revert "Revert "Revert "[redacted], unfinished""" a dozen changed files, a thousand added lines, five thousand deletions. Okay then.
|
|
#
?
Oct 20, 2023 13:30
|
|
|
I guess tomorrow you can look forward to Revert "Revert "Revert "Revert "[redacted], unfinished"""?
|
|
#
?
Oct 20, 2023 13:39
|
|
|
well-read undead posted:and however big you think the check will be, it will be slightly bigger a publicly-traded company had to restate their earnings because of an uncapped Datadog bill of tens of millions we migrated to self-hosted grafana and Prometheus and so forth this year because they are such assholes about pricing and we could staff a team and pay the cloud costs for way less. they were pretty shocked when we told them apparently
|
|
#
?
Oct 20, 2023 16:36
|
|
|
Antigravitas posted:Found this commit title today: 
|
|
#
?
Oct 20, 2023 19:39
|
|
|
Subjunctive posted:a publicly-traded company had to restate their earnings because of an uncapped Datadog bill of tens of millions if i had all the time in the world, and the people to do it, i would love to tell so many vendors to f off and bring more stuff in house. but the sheer scope of what we get from datadog is staggering, and would be super challenging to get parity with, and if we didn't all our developers would hate me man you think dealing with datadog is bad, try okta. 30% rate hike because "inflationary adjustment" lol motherfuckers
|
|
#
?
Oct 20, 2023 19:54
|
|
|
|
| # ? Apr 28, 2024 08:03 |
|
|
well-read undead posted:if i had all the time in the world, and the people to do it, i would love to tell so many vendors to f off and bring more stuff in house. but the sheer scope of what we get from datadog is staggering, and would be super challenging to get parity with, and if we didn't all our developers would hate me my cousin just told me a story from when he worked at a municipality they had some database running on a million+ a year contract with 3-4 on prem consultants, some ancient terminal system. one of the consultants was this 75 yo greybeard and it didnt look like he actually did much, so they cozied up to him and got to know more about the system, then set up 5 accounts to log in, run lookups & scrape everything into their own systems over a couple weeks when they dropped the vendor they were of course pissed and threatened fire & brimstone, but the contract was also old as poo poo and did not forbid it lol * i think the figure was like 5 million dkk in the 90s, so maybe 1.5 mil usd today?
|
|
#
?
Oct 20, 2023 20:02
|
|