Posts by cincodenada@cybre.space
(DIR) Post #AFaJQou8MotfQpXsxs by cincodenada@cybre.space
2022-01-19T02:51:34Z
0 likes, 0 repeats
@pry Oops it got worse: https://cybre.space/@cincodenada/107646796712147250
(DIR) Post #AFaJS8FtuDR9jnLudc by cincodenada@cybre.space
2022-01-19T02:51:49Z
0 likes, 0 repeats
@pry Guess what, it got worse: https://cybre.space/@cincodenada/107646796712147250
(DIR) Post #AGFHbjKs16UqUIXfSy by cincodenada@cybre.space
2022-02-07T20:57:47Z
1 likes, 1 repeats
Your regular reminder that NULL in databases doesn't mean "an empty value", it means "I literally don't know what this value is, it could be 6 or 6000 or -23, who's to say?"This difference is important, because "tell me all the people where wearing_hat is not true" - you might think you'll get records where wearing_hat is NULL, but you won't! They could be wearing a hat or not, NULL means you don't _know_ whether they're wearing a hat. The fact represented by wearing_hat could be true, even if the database value is NULL.And of course, if you ask if NULL = NULL, you don't get true, you get get NULL. Is Jody's hat-wearing status the same as Olaf's? If you don't know whether either of them are wearing a hat, you can't say that their hat-wearing status is the same.Much of this is the case for null values in various languages, but this is pretty universally true for databases, and trying to work with them without understanding this will cause you much pain!
(DIR) Post #AGFHbl7TNsy61KHzrU by cincodenada@cybre.space
2022-02-07T21:08:29Z
0 likes, 0 repeats
...a coworker telling me "nothing weird in the database, look I ran SELECT * FROM table WHERE field NOT IN('expected value', 'other expected value') and got nothing" and me saying "but what if you run SELECT * FROM table WHERE field IS NULL"Reader: there was something weird in the database, and it was 25 NULLs.Arguably, this state of the field is actually exactly the use case for NULL, but because this codebase was built from the perspective of "NULL is weird and breaks stuff", correctly setting it to NULL, well, breaks stuff. :blobpensiveleft:
(DIR) Post #AGFHbmyKUqqJlY1it6 by cincodenada@cybre.space
2022-02-07T21:11:31Z
0 likes, 0 repeats
I get very defensive of database designers and NULL because people are always like "NULL is the worst why would anyone ever make it work that way" and like, it's there for a reason, it fulfills a purpose, and it's solving a problem that's not easy to solve any other way.But also, yes it's a sharp edge that requires you to be aware of it, and that ties into learning curves and onboarding folks and "well if you just _know_ stuff you won't have this issue" is not a great answer. I don't know what the answer is.The obvious answer is "make NULL opt-in", but then you have to decide on an empty value besides NULL for every datatype, and that's not always obvious. What's the empty value for an image column? FALSE isn't always a safe bet for booleans. And if you try to pick a default value for a point on the earth's surface (which is a valid column type in PostGIS!) you end up with https://en.wikipedia.org/wiki/Null_Island🤷♂️
(DIR) Post #AIZ1YIDMtZnbmln9qS by cincodenada@cybre.space
2022-01-19T03:06:48Z
0 likes, 0 repeats
Update: it gets worse!!The placeholder text uses hyphens as separators, but the calendar control inserts slashes.So what happens if you literally type in the suggested placeholder text? Surely they wouldn't suggest an invalid format?If you guessed "of course they would" congratulations! You have the correct amount of faith in Atlassian's quality control! Blisteringly cursed software design, stratospheric levels of dgaf.Bonus: yes, the placeholder dates are in January *2012*, and I would not be surprised in the least if that means this code hasn't been updated in over a decade.
(DIR) Post #AIZ1YJGaz3yV34JFVw by cincodenada@cybre.space
2022-01-19T03:12:12Z
0 likes, 1 repeats
I don't do a lot of criticizing other people's software projects because I'm a developer and, like, software is hard! It's easy to screw stuff up!But the level of inconsistent shoddiness that Atlassian dependably puts out is organizational-level failure. This is not criticism of any of the developers involved, because they don't even have the ability to fail on this level.Atlassian is a company whose priorities are perfectly demonstrated by the fact that they spent engineering and support time implementing and defending a mandatory feature that autocorrects the capitalization of their brand names everywhere across their whole platform, a feature that serves literally no value to end-users: https://community.atlassian.com/t5/Jira-questions/Jira-gt-Deactivate-autocorrect-in-Jira-issues-comments-etc/qaq-p/1186807The end-user experience in their software is literally just not prioritized. Which unfortunately is a viable business model if the people you're selling the product to are not the ones using it.
(DIR) Post #AIZ1YNIVzrt3YBEq3s by cincodenada@cybre.space
2022-01-19T03:37:14Z
0 likes, 0 repeats
Welp, @avocatto couldn't just let this one lie, and observed that with two-digit years, you've got Y2K problems, right?You sure do. So I dug in: I figured they probably implemented the solution you use if for some reason HAVE to deal with two-digit dates (which they don't, but chose to anyway).You just pick a number, and assume any dates before that number are 20xx, and after are 19xx. For something like JIRA where you're not really recording historical data, you can probably get away with this. You're not gonna have tickets opened before, like, 1980, maybe pick 1970 cause it's the epoch, there, you're future proof for 50 years.So I went to test, and was surprised when 50 was interpreted as 1950...and a few astonished trials later: jimminy cricket on a bobsled, they chose nineteen-FORTY-TWO as the first year to assume is 19xx :blobthinkingeyes: :thonking: I really, really wish I were making this up because I am at a loss as to how you even begin to end up here. Stumped.
(DIR) Post #AIZ1YNJZvujnbTjgie by cincodenada@cybre.space
2022-01-19T03:14:11Z
0 likes, 0 repeats
(Apologies to folks who boosted or replied to the original version of this post: I went to redraft the *reply to this post* cause it had a pretty critical typo at the end, but redrafted this one instead, womp womp)
(DIR) Post #AIZ1YPFOkQa9b5nNTs by cincodenada@cybre.space
2022-01-19T03:40:52Z
0 likes, 0 repeats
My only guess, given the placeholder dates, is that this code was also written in 2012, and they decided that 30 years of buffer was good enough. Just bonkers decisionmaking.
(DIR) Post #AKAwlX8pTZHUXhWjyq by cincodenada@cybre.space
2022-06-05T01:51:27Z
0 likes, 0 repeats
Okay so I just went to create a multiple-choice poll and hot damn if that's not a totally undiscoverable interface, I was flummoxed for much longer than I should have been.(For those who haven't encountered this, you have just guess that clicking on the radio buttons next to the poll will change them to checkboxes :blobfacepalm:)
(DIR) Post #AKAwlXlT9uBoTXbbCy by cincodenada@cybre.space
2022-06-05T01:53:38Z
0 likes, 0 repeats
I'm grimly unsurprised that Eugene pooh-poohed this exact complaint already (because multiple-choice polls are "not a priority" and it would make the interface "too complex", and it's fine as "an easter egg"): https://github.com/mastodon/mastodon/issues/10873I have a bone to pick with a definition of "complex" that results in hiding basic features behind UX mechanisms that are impenetrable to the average user, creating a two-tiered experience where nerds like me who are steeped in How Computers Work get to have extra "easter egg" features like...allowing people to choose more than one poll option 🙄
(DIR) Post #AKAwlYBLbgtDlniGXo by cincodenada@cybre.space
2022-06-05T02:20:51Z
0 likes, 0 repeats
And in this instance it's especially silly, because the first two other interfaces I checked (Pinafore and Tusky) have entirely reasonable interfaces for this, both have dedicated and labelled "multiple choice" options, and neither have the radio buttons (because they're totally unnecessary for creating a poll!)
(DIR) Post #AKAwlcVLXHGPAz6IGu by cincodenada@cybre.space
2022-06-05T02:52:53Z
0 likes, 0 repeats
I don't have screenshots, but from looking at the source of glitch-soc it seems like they also have a reasonable interface for this. Looks like Hometown uses the stock Mastodon interface, but I just filed an issue to look into improving on that.In the spirit of not complaining without trying to fix it, I also intend to comment on the Mastodon issue, but right now my brain is tired of digging into this so that will have to wait for a bit.
(DIR) Post #AKTLwMlvSbb5bjDdvE by cincodenada@cybre.space
2022-06-13T21:52:35Z
1 likes, 0 repeats
Current gender: googling "cursive upper-case E"
(DIR) Post #ALdc8Ds4oetuHwKUPQ by cincodenada@cybre.space
2022-05-27T03:05:10Z
0 likes, 0 repeats
@bxl_forever @amapanda Okay I got a bit carried away here, but there's some thoughts below. In short, there's some precedent for colour=rainbow for the pride, but bi flag is harder, I didn't really find anything that had a colour pattern marked except for some bouys that have their own system. "colour=blue;purple;pink" with "note:colour=bi pride flag" is maybe the most conventional, but I'd like to find something more specific!Also worth noting that I'm not super active in tagging myself - I know the tools and reference materials fairly well, but haven't actually added that much data myself!
(DIR) Post #ANj8X7Gl6OWYuqxgtU by cincodenada@cybre.space
2022-09-18T03:43:11Z
0 likes, 0 repeats
Totally unbiased poll time...you write the following in a config file in a Unixy environment:include subdir/*Assuming this config language supports modular configs via including other config files in this manner, should dotfiles be included?#poll #configuration #unix #linux #globbing
(DIR) Post #ANxwzISSoNNZeAVTJA by cincodenada@cybre.space
2022-09-18T18:32:54Z
1 likes, 0 repeats
Tried to capture exactly how long I've been swimming in these open-source waters, and came up with "I was downloading Debian onto six CD's over dial-up with jigdo when Ubuntu was just a twinkle in Shuttleworth's eye" 😂It's far from ancient, and I didn't end up using it because I don't think pulling rank is helpful or useful, but I enjoyed it nonetheless. And a helpful reminder a lot of people have grown up in a totally different world than I have...something I have to keep reminding myself of more and more these days.Bonus early-aughts flavor: while waiting for my downloads, I printed Tux onto some labels with my parents' inkjet, then used a CD Stomper to apply them. Probably used OpenOffice 1.1? 🧓
(DIR) Post #ANxwzJ6WPRQDePFSkK by cincodenada@cybre.space
2022-09-18T19:04:49Z
0 likes, 0 repeats
Update: realized I still have those CD's and, most surprisingly, know where they are! And turns out it was a whole set. All told:- Debian 3.0r3 "Woody" (8 CD's)- Suse 10.0 (5 CD's) - Mandrake 10.1 (3 CD's) - Xandros Desktop 3.0 (1 CD)You can still see the sharpie underneath from before I put the labels on 😂
(DIR) Post #ANxwzJhkB3CDVqfBlQ by cincodenada@cybre.space
2022-09-18T19:12:00Z
1 likes, 0 repeats
Fun time capsule bonus: titles of the Known Issues from Xandros, verbatim:- RealPlayer does not launch when Sound Recorder is running- USB Floppy Drive detected as Removable Disk- Parallel port not present when printer not turned on during bootup- Double-clicking a .exe file in Xandros File Manager does not launch application with CrossOver- Open With's "Remember application association for this type of file" isn't respected for Windows apps.For those unfamiliar, CrossOver as I recall was WINE with a bunch of tweaks to make stuff "just work", and Xandros was a distro built around being a Linux that ran Windows apps.Kind of like, oh, a very early SteamOS/Proton situation I guess. The circle just keeps turning...