|
Cybernetic Vermin posted:the real vultures are (usually, don't know the precise system) those that pitch a complete replacement, implying huge cost savings, when the reality invariably turns out that it takes forever, disrupts the actually important activities, and certainly never saves a cent. I thought bitrot was a symptom of adopting too early, not holding on too late.
|
# ? Nov 13, 2019 12:39 |
|
|
# ? Jun 8, 2024 15:40 |
|
Cybernetic Vermin posted:the real vultures are (usually, don't know the precise system) those that pitch a complete replacement, implying huge cost savings, when the reality invariably turns out that it takes forever, disrupts the actually important activities, and certainly never saves a cent. there was some sci-fi book i read where the techs on some aging spaceship were called "code archeologists," and spent their whole lives digging through the trillions of lines of code in the ship's memory banks, looking for five emulators they could plug together to keep some ancient hack working that's the future assuming industrial civilization doesn't end in the next century imo
|
# ? Nov 14, 2019 04:15 |
|
animist posted:there was some sci-fi book i read where the techs on some aging spaceship were called "code archeologists," and spent their whole lives digging through the trillions of lines of code in the ship's memory banks, looking for five emulators they could plug together to keep some ancient hack working vernor vinge, a deepness in the sky
|
# ? Nov 14, 2019 04:32 |
|
Notorious b.s.d. posted:vernor vinge, a deepness in the sky love that series
|
# ? Nov 14, 2019 04:41 |
|
animist posted:there was some sci-fi book i read where the techs on some aging spaceship were called "code archeologists," and spent their whole lives digging through the trillions of lines of code in the ship's memory banks, looking for five emulators they could plug together to keep some ancient hack working mystes fucked around with this message at 04:44 on Nov 14, 2019 |
# ? Nov 14, 2019 04:41 |
|
mystes posted:I think the chance of a future where humans are janitoring ancient code on spaceships is pretty much zero, which is about one percent less than our chance of surviving in the first place. maybe not on spaceships but we will definitely be maintaining lovely cobbled together layers of weird code for centuries (there is code in prod that's like 50 years old at this point)
|
# ? Nov 14, 2019 06:03 |
|
he was a Sergeant at Arms Programmer Archeologist, studying ancient software was his military duty of course the Blight was released by some arrogant techbros that thought JavaScript was close to the metal
|
# ? Nov 14, 2019 06:08 |
|
oh and the Blight’s ubiquitous slaves were some garbage internet of things that absolutely nobody understood
|
# ? Nov 14, 2019 06:09 |
|
Krankenstyle posted:maybe not on spaceships but we will definitely be maintaining lovely cobbled together layers of weird code for centuries (there is code in prod that's like 50 years old at this point) it would be suicide to let a programmer on any sort of generational ship since they would be compelled to rewrite the whole stack in whatever the latest coding fad they got transmitted from earth. it would be for the best if the entire first generation instilled a religious dogma to never ever try and understand the code that runs the ship
|
# ? Nov 14, 2019 06:10 |
|
like, we got templeOS here on earth, where there's vast access to tons of programming resources and mental health providers. imagine the hosed up poo poo a programmer cloistered on a ship light years from earth would come up with
|
# ? Nov 14, 2019 06:12 |
|
Krankenstyle posted:maybe not on spaceships but we will definitely be maintaining lovely cobbled together layers of weird code for centuries (there is code in prod that's like 50 years old at this point) the ibm system 360 (1964) was backwards compatible with a bunch of 1950s machines, so there is probably still code out there that has celebrated its 60th birthday also cobol was first published in 1960 there's a ton of old poo poo out there
|
# ? Nov 14, 2019 06:14 |
|
Plank Walker posted:it would be suicide to let a programmer on any sort of generational ship since they would be compelled to rewrite the whole stack in whatever the latest coding fad they got transmitted from earth. it would be for the best if the entire first generation instilled a religious dogma to never ever try and understand the code that runs the ship you are suffering from the delusion that the code that runs a generation ship would actually work
|
# ? Nov 14, 2019 06:19 |
|
smdh if you don't want to rewrite your generational ship's power management system in Javascript so it can run anywhere
|
# ? Nov 14, 2019 08:11 |
|
Notorious b.s.d. posted:the ibm system 360 (1964) was backwards compatible with a bunch of 1950s machines, so there is probably still code out there that has celebrated its 60th birthday tbh its kinda impressive tho probably the earlier code is pretty robust since it presumably wasnt written by complete idiots
|
# ? Nov 14, 2019 09:09 |
|
Krankenstyle posted:tho probably the earlier code is pretty robust since it presumably wasnt written by complete idiots it probably was written by idiots. or at least bright people who had no idea what they were doing. it’s probably robust because folks have been keeping it limping along for decades
|
# ? Nov 14, 2019 09:18 |
|
i mean their standards for correctness were probably better than they are now (like the apollo code) but yeah probably thats fair
|
# ? Nov 14, 2019 12:12 |
|
the jpl spent an obscene amount of money achieving their level of code quality because it was really loving important lol if you think the average ibm customer from that era had a comparable desire and budget for robust code
|
# ? Nov 14, 2019 12:37 |
|
alright apollo was a bad example, i just like to imagine that programs werent always dogshit
|
# ? Nov 14, 2019 13:04 |
|
mostly old software is more likely to be correct because of it wasn't there would have been time intervening to fix it. it also accumulates cruft with the fixes, but as we're never getting to a state of all software being new and correct one should have a healthy respect for things that get work done.
|
# ? Nov 14, 2019 13:25 |
|
Krankenstyle posted:i mean their standards for correctness were probably better than they are now (like the apollo code) but yeah probably thats fair there is still code like this, but the ratio of code like this to code not like this has gotten even more unfavorable since the apollo era
|
# ? Nov 14, 2019 15:26 |
|
Old software was/is also a lot simpler (by necessity) and had much more stringent requirements on input. It's perfectly fine to read a line into the COBOL equivalent of a char[80] buffer when the input comes from punched cards where lines physically cannot be longer.
|
# ? Nov 14, 2019 15:29 |
|
Athas posted:Old software was/is also a lot simpler (by necessity) and had much more stringent requirements on input. It's perfectly fine to read a line into the COBOL equivalent of a char[80] buffer when the input comes from punched cards where lines physically cannot be longer. cobol was also designed by people who understood the dangers of types like a char[80] buffer so that is physically impossible in cobol afaik when you read data into a string in cobol you can optionally provide an event handler for "what if it overflows?" but you can't accidentally overwrite other poo poo i am not a cobol programmer though this is just poo poo i barely remember from touching it once in university
|
# ? Nov 14, 2019 17:40 |
|
Bloody posted:there is still code like this, but the ratio of code like this to code not like this has gotten even more unfavorable since the apollo era i've been reading a lot of stuff about formally verified systems for a class, and I keep reading stuff like quote:A commercial example concerns key modules of a preemptive OS kernel, the μC/OS-II. Modules verified include the scheduler, interrupt handlers, and message queues. 1.3k lines of C were proven using 216k lines of Coq. It took four person years to develop the framework, one-person year to prove the first module, and then the remaining modules, around 900 lines of C, took six person-months.
|
# ? Nov 14, 2019 17:56 |
|
cobol i/o is all about fixed-size records, but yes, it is a significantly higher-level language than c and you would really need to work in order to get it to scribble over memory. the flip side is that cobol is pretty go-like in the sense that user-defined data structures are limited to basically structs, and otherwise you get global tables and global indexed tables and that's basically it
|
# ? Nov 14, 2019 18:04 |
|
rjmccall posted:cobol i/o is all about fixed-size records, but yes, it is a significantly higher-level language than c and you would really need to work in order to get it to scribble over memory. the flip side is that cobol is pretty go-like in the sense that user-defined data structures are limited to basically structs, and otherwise you get global tables and global indexed tables and that's basically it unlike golang nobody marketed cobol for "systems programming"
|
# ? Nov 14, 2019 18:06 |
|
Notorious b.s.d. posted:unlike golang nobody marketed cobol for "systems programming" obligatory: cobol has had generics for almost a decade
|
# ? Nov 14, 2019 21:22 |
|
also really old codebases are solid for the same reason Roman bridges are solid: because all the lovely ones had to be replaced, and only the best didn't. not because Roman / cobol engineers were supermen
|
# ? Nov 14, 2019 21:37 |
|
Huh, it's survivor bias again, never thought of it in that light again
|
# ? Nov 14, 2019 22:45 |
|
dick traceroute posted:Huh, it's survivor bias again, never thought of it in that light again whenever old poo poo seems of higher quality than recent poo poo it's usually this
|
# ? Nov 14, 2019 22:55 |
|
Planned obsolescence is also a factor
|
# ? Nov 14, 2019 23:02 |
|
or maybe everyone was better but only the most boring poo poo that no one woild ever want to rewrite is the remaining stuff
|
# ? Nov 14, 2019 23:09 |
|
Well I think it's safe to say the 12 year old shitpile VB winforms + oracle stack I just left at oldjob was unintentional
|
# ? Nov 14, 2019 23:10 |
|
we have way better tooling now though than anyone ever had back in the 60's, both for testing and for static analysis. only certain industries bother though and verification tools like e.g. astrée are usually pretty expensive
|
# ? Nov 15, 2019 00:06 |
|
TheFluff posted:we have way better tooling now though than anyone ever had back in the 60's, both for testing and for static analysis. only certain industries bother though and verification tools like e.g. astrée are usually pretty expensive you can beta test in the future to realize gains from decreasing costs of testing tools.
|
# ? Nov 15, 2019 00:31 |
|
Vomik posted:or maybe everyone was better but only the most boring poo poo that no one woild ever want to rewrite is the remaining stuff 'to err is human but to really screw things up takes a computer' was a saying for a reason
|
# ? Nov 15, 2019 02:00 |
|
animist posted:216k lines of Coq
|
# ? Nov 16, 2019 00:32 |
|
lutha pls go back to your late medieval avatar, i dont like this early "modern" avatar at all.
|
# ? Nov 16, 2019 01:43 |
|
somebody bought it for me and I haven't thought of what hi-larious new av gimmick i should try next
|
# ? Nov 16, 2019 02:52 |
|
Lutha Mahtin posted:somebody bought it for me and I haven't thought of what hi-larious new av gimmick i should try next ideas: - martin luther but blingee - 95 theses but on a computer screen(!) - lets get weird: danny devito in a 1500s getup and a bag of beers (ill chip in if he asks for cash)
|
# ? Nov 16, 2019 03:10 |
|
|
# ? Jun 8, 2024 15:40 |
|
Lutha Mahtin posted:somebody bought it for me and I haven't thought of what hi-larious new av gimmick i should try next the rms / luther martin connections are pretty strong though
of course only one of them is famous for defending a dead sex pest so, there's that
|
# ? Nov 16, 2019 03:40 |