Programming as a career

SQL rules

I’ve been playing around a lot in R recently. I like that it’s more of a really powerful calculator than a programming language.

Regex'es are painful to work with, no doubt, but there's often not a simpler alternative to accomplish what you can do with them. Actually one thing that I don't like about SQL (despite it's being mostly awesome) is the lack of regexes.

SQL wuz here ‘19

I’m a bad person who does SQL irregularly and rather than remember how to do a partition statement the other week, I once again wrote a query in LinqPad and had it generate a query for me

It never ceases to amaze me how much “mission critical” shit is written in toy languages designed to be intentionally “dumber” than Real Computer Scientist code

― El Tomboto, Friday, July 12, 2019 2:39 PM (yesterday) bookmarkflaglink

I’ve known a few people who worked on code for fighter jet stuff and it was all, by regulation, done in Ada. I think it’s less dumb than incredibly explicit and lacking abstractions that would make fuck-ups easier

Apparently the rules have changed and the F-35 has been programmed in C++, which makes sense considering how much of a boondoggle that entire project is

Yeah I heard ADA is used in aeronautics and shit because it’s really really hard to fuck up (there’s a word in programming language concepts for this, extra safe or something )

seeing some dumb fite on Twitter about '10x engineers' with no idea what means

developers' inability to use SQL these days does get a bit infuriating sometimes.

my colleague added a page to the CMS just to add values to a lookup table the other day then deleted the page when he was done, because he can't write an INSERT statement. but he is completely hopeless and hopefully will get fired soon

I'm not sure why SQL doesn't support regex out of the box really. it's completely trivial to add it to SQL Server with a C# CLR function, so why they don't just include it as standard I don't know. I think Postgres might support them?

Postgres has built in regex support yeah

& mysql.

and Oracle, I'm pretty sure (I use regexp_like and regexp_replace on a p. much daily basis)

I like SQL but I'm glad my queries are mostly for one-off reports and aren't called regularly or made available for Joe Public to dial up at will because the more interesting ones (and some of the less interesting ones) get really fucking slow

this is partly because the tables don't always join in a sensible manner and useful fields aren't indexed but also I don't know what I'm doing and can't read an "explain plan"

so, I like SQL, but should probably not be allowed to write queries used in the back-end of anything busy

...I say I "like" SQL, I also swear at it a lot and find it ugly and etc

sometimes I have even been known to miss features from our previous database, a multi-value Pick database which felt like it was straight out of the 70s (though the iteration I was using was apparently out of the, ooh, 80s)

(the 2nd link there is the only reference material I could find anywhere apart from the original 80s manual, which was falling apart, jealously guarded on the boss's desk, and also borderline unreadable - so luckily that site there is actually very good and rescued me from many an impending disaster)

I was about to say something that I thought was interesting and constructive, but instead read that Wiki and found that OS is named after Dick Pick.

I believe ILX has created a repository for this kind of stuff? Great Real Names ?

a standard regex query in sql sounds neat

the problem is probably related to a bunch of vendors coming up with extra shitty versions of useful features, fighting over what should be standard

a coworker did a presentation explaining how oracle’s xml querying (which was probably an insanely bad idea anyway) was so fucked they actually tore it out of a newer version and retooled it

generally if you want to do document storage or mongo stuff, just... use a document storage db. relational databases don’t need to include every damn thing and doing so just encourages people to do very bad things

I can see not wanting to include regexes out of the box, since it's possible for an unlucky combo of regex and input string to max out CPU and bring the server to its knees. I think it was a bad regex that brought Cloudflare offline a few weeks ago.

yep that was a great postmortem.

who decided that percentages in en-US should have a space between the number and the percentage sign and can I shake them until their teeth rattle?

that thing where it's easier to rewrite the code than it is to debug it.

(somewhere in ~3000 packets the data is getting out of sync but this only manifests as an error in the last packet. just can't nail it down)

((there's a peculiar case where you've got just enough data to go into a fixed-length packet but when you've added the extra 2 bytes to say that this is the last packet you no longer have the space for the all data, so this ends up being the last-but-one packet and you have to add another packet containing a single byte))

fwiw, the second attempt at the code is much neater.

as long as you're sure the original code didn't have obscure and undocumented but vital side effects...

it's ALL obscure and undocumented side effects.

(it's not, it's new, non-production code, something i wrote 3 weeks ago and have spent the last two debugging. file in, slightly modified file out. unfortunately the output is fixed-length packets and some of the modifications add / remove one or two bytes from some of the packets, meaning all the rest of the data has to shuffle down and new packets may be created)

code review. github tells me 84 files have changed. +769 -451 lines

in those files there are *4* comments, all similar to this (but with different numbers, 1-4)

    generateNoof(malUnits),  // #1

the "84 files changed" aspect of that is really what scares me the most

we recently went on a big code formatting kick and committed basically our entire codebase which sucks because now the 'show annotations/history' feature in RAD points every line to that one revision cuz it replaced all the tabs with spaces

my last job the codebase had a "tabs-to-spaces" commit that messed up git blames but it's possible to skip over it/reblame starting before that commit in various ways

cf git blame --ignore-rev

it's a refactor of some classes into smaller classes. so all the test classes need refactoring too. and then some test utility things have been changed to static methods so everything that used them have also changed.

(it might also be that the 84 is for the entire pull request whereas i've only looked at the first commit so far)

and some default formatter keeps dicking wih comments so you end up with stuff like

// very long comment that would be fine all on one line but the formatter
// disagrees

wait, is RAD whatI think it is? If so, condolences

I finish bootcamp in two weeks, gang! :D

Congrats...was this an online bootcamp? How was it affected by covid?

Which one did you do? Was it a Rails one, or JS, or something else?

I did ReactJS!

I went to an actual campus six days a week until the COVID lockdown

my last job the codebase had a "tabs-to-spaces" commit that messed up git blames but it's possible to skip over it/reblame starting before that commit in various ways

― silby, Tuesday, April 28, 2020 12:17 PM (four days ago) bookmarkflaglink

cf git blame --ignore-rev

― silby, Tuesday, April 28, 2020 12:17 PM (four days ago)

Oh nice! https://www.moxio.com/blog/43/ignoring-bulk-change-commits-with-git-blame

Cool

lmao I didn't even realize that was a new feature, I perhaps haven't actually used that yet. idk whatever.

anyway, condolulations Whiney

Thanks!

definitely going to make use of --ignore-rev from now on but it'll be a pain to set that up for every giant squash merge.

squash mergers being fucking savages is the only programming hill i'm willing to die on

sorry whiney but you have to turn in your cool badge at the security desk now, prepare to be bullied, NERD

breh if i'm working on a giant project i'm definitely squash merging my fuckton of commits into 1 for the sake of keeping the master branch clean

smh if you aren’t interactively rebasing your feature branches into a glittering necklace of logical and well-described commits.

sorry whiney but you have to turn in your cool badge at the security desk now, prepare to be bullied, NERD

― j., Sunday, May 3, 2020 2:50 PM (two hours ago) bookmarkflaglink

going from "geek" to "nerd" will be a difficult transition but not an impossible one

It's been fun meeting people who have no idea what noise music is!

silby otm.

Just you wait, Whiney.

having a true commit history is infinitely more useful than having a clean master branch. also you should be doing what silby said

always said i'd leave ilx after stet adds inline manscaped adds

Congratulations whiney!

I did ReactJS!

Welcome fellow ReactJS dev!

"I did ReactJS!"

How is the 40 hour course on Udemy, anyone know?