|
what I said to the offshore team: "the grid result need to be exported to excel serverside so we need to look at how the legacy methods work because they're bad. it would be nice if we generated actual xlsx files but I think it might need Excel installed on the server so might not work. also data should be returned as a byte stream to the browser. " what came back: literally the exact opposite. serverside com interop that needs a local excel install so fails because it's not installed on Web hosts and also memory leaked excel instances everywhere, storing temp files on the server and not deleting them and my favourite, returning a *path to a temp file on the server* as a response to the Web request that is then sent back to the server from the browser as another request to a controller that then just returns whatever file is at that path with zero access control i don't know why I bother.
|
# ? Sep 15, 2017 19:01 |
|
|
# ? May 10, 2024 07:06 |
|
oh poo poo I forgot that they also duplicated all of the data methods as well for the sole purpose of adding a single hard coded bit flag parameter to the procedure calls instead of just making it an optional parameter on the main method i replaced all their code
|
# ? Sep 15, 2017 19:07 |
|
Powerful Two-Hander posted:offshore team
|
# ? Sep 15, 2017 19:26 |
|
wisdom Kind of related but a shout out gently caress you to Microsoft on COM interop which is still a pile of poo poo that doesn't shut down instances properly unless you janitor them manually
|
# ? Sep 15, 2017 20:25 |
|
Luigi Thirty posted:also according to the DSP, the absolute value of signed zero ($80000000) is signed zero ($80000000) errata: accidentally implemented a hardware random number generator instead of a processor core workaround:
|
# ? Sep 15, 2017 21:06 |
|
we have a contracted team in south america and they're actually really good. best contractors ive ever worked with
|
# ? Sep 15, 2017 21:42 |
|
Powerful Two-Hander posted:it would be nice if we generated actual xlsx files but I think it might need Excel installed on the server so might not work actually there are decent libraries for making xlsx files without excel. I've used the python one with good results.
|
# ? Sep 15, 2017 22:04 |
|
Soricidus posted:actually there are decent libraries for making xlsx files without excel. I've used the python one with good results.
|
# ? Sep 15, 2017 22:05 |
|
FamDav posted:jetbrains had a problem: they had to hold features because they needed to make upgrades worthwhile, and they would rather release features when ready and adopt an evergreen model. they wanted to move to a straight subscription model, but customers who liked having real ownership of the yearly release were mad, so they made it so you could get ownership of a release if you paid for a year. everybody is super happy about it and got what they wanted
|
# ? Sep 15, 2017 22:08 |
|
anthonypants posted:did you know??? .xlsx is actually a .zip i've encountered a file that claimed to be xlsx that was actually an xml file. the only way i found to actually deal with it was open it in excel (with a bunch of warnings) and then actually save it as a real xlsx
|
# ? Sep 15, 2017 22:22 |
|
gonadic io posted:i've encountered a file that claimed to be xlsx that was actually an xml file. the only way i found to actually deal with it was open it in excel (with a bunch of warnings) and then actually save it as a real xlsx
|
# ? Sep 15, 2017 22:51 |
|
gonadic io posted:i've encountered a file that claimed to be xlsx that was actually an xml file. the only way i found to actually deal with it was open it in excel (with a bunch of warnings) and then actually save it as a real xlsx Oh, God, was that Excel 2003? Single XML file (XMLSS), should be a .xml, not actually readable by basically anything other than Excel and certainly not by the Apache POI project that handles all the other Excel formats?
|
# ? Sep 15, 2017 22:52 |
|
ulmont posted:Oh, God, was that Excel 2003? Single XML file (XMLSS), should be a .xml, not actually readable by basically anything other than Excel and certainly not by the Apache POI project that handles all the other Excel formats? quite possibly. i just made the person giving it to me open-and-resave it so i could use poi
|
# ? Sep 15, 2017 23:04 |
|
Soricidus posted:actually there are decent libraries for making xlsx files without excel. I've used the python one with good results. yeah i suggested this and it just got blank silence and i couldn't be bothered to wade through 800 implementations to find a non-poo poo one myself so i just dumped the data to .csv and converted it to a bytesream. worksforme bug closed. anthonypants posted:did you know??? .xlsx is actually a .zip i keep telling people this and they literally do not believe me i also encountered this bullshit: if you load a .csv file into excel and the first column name is "ID" it will fail to parse it because excel thinks every file beginning with "ID" is a "SYLK" file and then throws a "cannot parse this file" error so you have to add a fix to change it to "id" Powerful Two-Hander fucked around with this message at 23:29 on Sep 15, 2017 |
# ? Sep 15, 2017 23:26 |
|
anthonypants posted:did you know??? .xlsx is actually a .zip
|
# ? Sep 16, 2017 00:27 |
|
Powerful Two-Hander posted:i also encountered this bullshit: if you load a .csv file into excel and the first column name is "ID" it will fail to parse it because excel thinks every file beginning with "ID" is a "SYLK" file and then throws a "cannot parse this file" error so you have to add a fix to change it to "id"
|
# ? Sep 16, 2017 00:28 |
|
i swear that MS is maintaing some kind of "old school knowledge" program to ensure that some day, somehow, when somebody says "what is this 'cannot parse sylk file error'?" i can say "heh, actually..." see also the classic "dtsx file parsing only looks at the first 1000 rows so will gently caress you over if you have a longer data type in row 1001"
|
# ? Sep 16, 2017 00:44 |
|
Sweevo posted:errata: accidentally implemented a hardware random number generator instead of a processor core essentially yes I’m currently porting my C matrix math library over to the Jag DSP because I am literally insane
|
# ? Sep 16, 2017 01:42 |
|
Creating the new style xlsx files is ok with some of the libraries, although there's a load of boilerplate. God help you if you have to parse one though (especially user created. Just use csv or the far superior tsv instead
|
# ? Sep 16, 2017 10:14 |
quiggy posted:hey what's a good issue tracker these days Trello
|
|
# ? Sep 16, 2017 20:33 |
|
PokeJoe posted:Trello they picked Bugzilla
|
# ? Sep 16, 2017 20:51 |
|
Jaguar RISC manual posted:G_PC: The GPU program counter register. Changing this while the GPU is running is probably not a good idea. today’s adventure is trying to figure out how to tell the GPU to start executing a program at a specific offset so I can load my whole matrix math library into memory and call add/subtract/etc like subroutines the RISC CPUs are something like 10-12x faster than the 68k in matrix operations taking the doubled clock speed and near 1 cycle per instruction timing into account, and you can do the operations in parallel with the CPU.
|
# ? Sep 17, 2017 00:35 |
|
THEN WHY DID YOU PUT IT IN THE MANUAL
|
# ? Sep 17, 2017 04:04 |
|
Luigi Thirty posted:
optimism
|
# ? Sep 17, 2017 04:17 |
|
Luigi Thirty posted:
courage
|
# ? Sep 17, 2017 05:10 |
|
carry on then posted:optimism in reality this means there's a whole chapter of "DON'T TOUCH THESE REGISTERS OR YOU'LL BLOW UP YOUR TV" video settings
|
# ? Sep 17, 2017 07:21 |
|
Luigi Thirty posted:
because they didn't wanna re-run the layout tool lol
|
# ? Sep 17, 2017 07:27 |
|
Luigi Thirty posted:
that doesn't disprove their claims analog video output is a nightmare
|
# ? Sep 17, 2017 07:27 |
|
Luigi Thirty posted:
I mean, not so much the chip itself, but the jag did indeed become the backbone of many products that were definitely unforeseen A toothbrush holder, for example.
|
# ? Sep 17, 2017 14:17 |
|
Notorious b.s.d. posted:that doesn't disprove their claims presumably vaguely related, I've always assumed HDMI is a hilarious nightmare to implement, but I've never bothered to attempt to verify that assumption. so if you or anyone has a handy link detailing, idk, some relevant insanity, I'd read it
|
# ? Sep 17, 2017 16:41 |
|
pokeyman posted:presumably vaguely related, I've always assumed HDMI is a hilarious nightmare to implement, but I've never bothered to attempt to verify that assumption. so if you or anyone has a handy link detailing, idk, some relevant insanity, I'd read it Not really. There is a well defined standard and it's digital. Most cpus have it built in and you just have to run the traces. Ezpz. Auto-negotiation is part of the standard as well, so resolution and hertz rate are easy enough to decipher as well.
|
# ? Sep 17, 2017 17:02 |
|
i think upnp is the standard that's just complete unimplementable garbage
|
# ? Sep 17, 2017 17:18 |
|
Luigi Thirty posted:also according to the DSP, the absolute value of signed zero ($80000000) is signed zero ($80000000) This is what you get from pretty much all 2s-complement types. $80000000 is -2147483648, and since 32-bit integers only go up to 2147483647, adding 1 to that gives you -2147483648 again.
|
# ? Sep 17, 2017 20:52 |
|
Wheany posted:i think upnp is the standard that's just complete unimplementable garbage ive always wondered that about pcmcia and/or expresscard
|
# ? Sep 18, 2017 02:11 |
|
ratbert90 posted:Not really. There is a well defined standard and it's digital. Most cpus have it built in and you just have to run the traces. Ezpz. Auto-negotiation is part of the standard as well, so resolution and hertz rate are easy enough to decipher as well. that's cool. happy to be wrong!
|
# ? Sep 18, 2017 03:37 |
|
pokeyman posted:presumably vaguely related, I've always assumed HDMI is a hilarious nightmare to implement, but I've never bothered to attempt to verify that assumption. so if you or anyone has a handy link detailing, idk, some relevant insanity, I'd read it DVI and DVI are honestly not so bad, they're just transports for bitstreams, you need a transceiver but otherwise your video hardware will probably generate a compatible bitstream easily (and if you're using an FPGA there'll probably be a functional block that does so that you can just drop in)
|
# ? Sep 18, 2017 06:02 |
|
Zemyla posted:This is what you get from pretty much all 2s-complement types. $80000000 is -2147483648, and since 32-bit integers only go up to 2147483647, adding 1 to that gives you -2147483648 again. What he's describing sounds more like a representation with an actual sign bit, tho, like IEEE floating point for eg?
|
# ? Sep 18, 2017 13:13 |
|
i've been programming various poo poo for 15 years and i just realized literally today that underscores are magic characters in sql LIKE-statements
|
# ? Sep 19, 2017 00:52 |
i just found out that our mongo data is converted from xmls (half-assedly, explaining schema inconsistencies), and when i asked about that to confirm the fact, the "lead database guy replied" THAT THEY ARE CONVERTING XML RESPONSES TO JSON SO THEY COULD BE STORES IN A DATABASE
|
|
# ? Sep 19, 2017 10:14 |
|
|
# ? May 10, 2024 07:06 |
words can't explain in english the extent of ёбаный стыд this speaks of given that we are running postgre servers amongst other things
|
|
# ? Sep 19, 2017 10:15 |