|
its cool that C has a million build systems and they are all awful
|
# ? Feb 17, 2020 21:01 |
|
|
# ? Apr 28, 2024 17:43 |
|
today i used nvidia's extensions to visual studio to build some cuda, so: agreed
|
# ? Feb 17, 2020 21:02 |
|
Nomnom Cookie posted:within the last year i've run autotools builds that failed on "unnecessary" "legacy" checks because apple is doing exciting innovative things with breaking toolchains
|
# ? Feb 17, 2020 21:56 |
|
Zlodo posted:why would you possibly want to compile anything for the most willfully non-standard linux like that nobody uses i don't want to compile, but its what i get paid for, so
|
# ? Feb 18, 2020 00:06 |
|
Nomnom Cookie posted:phk writes entertainingly and with conviction, but he's also a unix philosophy dipshit who values architectural purity over working systems. his criticisms in that article boil down to "autotools is ugly"--true, but not important--and "it runs checks that I don't think are necessary because all the world is linux now (also freebsd though because i am a freebsd dev)". within the last year i've run autotools builds that failed on "unnecessary" "legacy" checks because apple is doing exciting innovative things with breaking toolchains I did an auto tools configure today and it was checking for loving stdlib.h. then I tried to build the thing and it turned out to have a bunch of actual compatibility issues that auto tools didn’t check for. making me wonder what the point was supposed to be. at least I didn’t notice it checking for fortran compilers this time. maybe they finally changed it to only do that if it’s actually maybe going to compile some fortran?
|
# ? Feb 18, 2020 03:09 |
|
Soricidus posted:I did an auto tools configure today and it was checking for loving stdlib.h. then I tried to build the thing and it turned out to have a bunch of actual compatibility issues that auto tools didn’t check for. making me wonder what the point was supposed to be. autotools can check for all the fortran compilers it wants imo. it's just not worth caring about
|
# ? Feb 18, 2020 04:34 |
|
fartron
|
# ? Feb 18, 2020 04:51 |
|
The best build system for C Lang’s is: - meson - cmake - bash scripts - calling gcc by hand - gradle - auto tools - killing yourself - bazel
|
# ? Feb 18, 2020 05:09 |
|
gcc -o fart *.c imo
|
# ? Feb 18, 2020 08:16 |
|
https://twitter.com/Foone/status/1229641258370355200
|
# ? Feb 18, 2020 17:52 |
|
lmao
|
# ? Feb 18, 2020 19:46 |
|
I send apologies to University of Texas-ans, you may get to suffer my next project, wayfinding. Here are my golden rules that precisely every system I have ever seen break: Rule 1: make the route obvious, Rule 2: detail should be on the route not minor, irrelevant content. So many maps like to highlight cabinetry, wall thickness and whatever else the cartographer is masterbating over. Rule 3: local location should be highlighted at center-bottom like every normal maps service, sticking random point on the screen is just confusing. Oh gently caress.
|
# ? Feb 18, 2020 23:09 |
|
c tp s: wpf is cool and good
|
# ? Feb 18, 2020 23:37 |
|
yah it's also v microsoft to try and replace it like 3 times with demonstrably worse frameworks
|
# ? Feb 18, 2020 23:39 |
|
what did they actually try to replace it with? isnt uwp just wpf but on dotnet core or something?
|
# ? Feb 19, 2020 02:00 |
|
Bloody posted:what did they actually try to replace it with? isnt uwp just wpf but on dotnet core or something? yes until no (note: my experience is years out of date but I suspect little has changed)
|
# ? Feb 19, 2020 03:26 |
|
ctps: poorly reinventing a tcplike network stack on top of modbus serial
|
# ? Feb 19, 2020 15:23 |
|
why not simply create tcp/modbus
|
# ? Feb 19, 2020 21:14 |
|
lol you can do modbus over tcp i like where this is headed
|
# ? Feb 19, 2020 21:14 |
|
use SLIP
|
# ? Feb 19, 2020 21:17 |
|
That PLC library which had the gaping security hole in linux for years I posted about before was libmodbus. I feel like it's enough time since that I can say that. So depending on how old your libraries are it might be safer to implement it yourself. e: or just compile the library from source it's not that complicated, it's been fixed for about 2 years now e2: I've also implemented kernel regmap interface/library for modbus over various things, it's not in-tree though Private Speech fucked around with this message at 21:35 on Feb 19, 2020 |
# ? Feb 19, 2020 21:19 |
|
Bloody posted:why not simply create tcp/modbus Bloody posted:lol you can do modbus over tcp that would require more expensive hardware than we have that im trying to talk to. it's rs485 only and im just glad it's not ascii Private Speech posted:That PLC library which had the gaping security hole in linux for years I posted about before was libmodbus. binding to libmodbus from rust was one of the options i was considering so uhhhh
|
# ? Feb 19, 2020 21:52 |
|
gonadic io posted:that would require more expensive hardware than we have that im trying to talk to. it's rs485 only and im just glad it's not ascii just compile from source, modbus is really simple and I don't think the library is all that complex, mainly the tcp stuff has some requirements but still for rs485 it shouldn't even really matter security-wise anyway, since, you know, you're already kinda hosed if someone is splicing into your wires, TCP could be more problematic e: or implement it yourself the protocol is trivial and at least you'll know that any buffer overflows and opening up ports on *.*.*.* is your own fault (not joking) Private Speech fucked around with this message at 22:22 on Feb 19, 2020 |
# ? Feb 19, 2020 22:17 |
|
iot bitches on that subject I like this talk b/c yes its really accurate still: Black Hat 2013 - Exploiting Network Surveillance Cameras Like a Hollywood Hacker I've literally personally had the "vendor sends the password in plaintext as part of a proprietary login packet" happen to me. AND they were mildly upset that we really insisted they fix it quick
|
# ? Feb 19, 2020 22:31 |
|
Private Speech posted:just compile from source, modbus is really simple and I don't think the library is all that complex, mainly the tcp stuff has some requirements but still yeah implementing it myself is reasonably likely, rust should help with the buffer overflows and there will be no ports involved whatsoever.
|
# ? Feb 19, 2020 22:32 |
|
i've only just started on the spring boot docs and i've already seen several mentions of things being automatically set up if they're available on the classpath getting a bit scared
|
# ? Feb 19, 2020 22:48 |
|
spring is magic
|
# ? Feb 19, 2020 23:29 |
|
also @Bean and @Autowired everywhere
|
# ? Feb 19, 2020 23:29 |
|
its not magic its just a terrifying abuse of reflection
|
# ? Feb 19, 2020 23:31 |
|
Bloody posted:its not magic its just a terrifying abuse of reflection the main thing I hate about reflection is how much of a pain it is for code suggesters like intellisense and the like by which I mean the best they usually do is just giving you a massive list and telling you to go e: this is even better in python where suggested argument types are rather important in any sizable codebase Private Speech fucked around with this message at 23:41 on Feb 19, 2020 |
# ? Feb 19, 2020 23:39 |
|
Spring is horrifying and I wrote my own URL router thing atop guice just to get away from it. I should probably actually add some docs to that github repo at some point. More to the point I'm not sure I'd start a new Java project these days if only because all the db access libraries in wide use are godawful Yes that includes jooq Why yes I would like my build system to connect to a db instance so that it can dump out the schema for codegen purposes
|
# ? Feb 19, 2020 23:57 |
|
you can use jooq to dump the schema to an xml file that can be used to generate the code from it instead of spinning up a db instance. we have our CI spin up a docker container, then run all the migrations, then run the codegen. it's pretty easy jooq owns
|
# ? Feb 20, 2020 00:40 |
|
CRIP EATIN BREAD posted:we have our CI spin up a docker container, then run all the migrations, then run the codegen. our ci does that too and its godawful Sapozhnik posted:Spring is horrifying and I wrote my own URL router thing atop guice just to get away from it. spring isn't the horrifying part of this post
|
# ? Feb 20, 2020 04:15 |
|
suffix posted:i've only just started on the spring boot docs and i've already seen several mentions of things being automatically set up if they're available on the classpath this is how all of spring boot works and it sucks rear end op. it's not java you're writing anymore, it's spring boot
|
# ? Feb 20, 2020 08:22 |
|
i dislike annotation-heavy java libraries because all they really accomplish is turning compile errors into runtime errors
|
# ? Feb 20, 2020 08:40 |
|
Progressive JPEG posted:i dislike annotation-heavy java libraries because all they really accomplish is turning compile errors into runtime errors or, in the case of lombok, turning compile errors into internal compiler errors
|
# ? Feb 20, 2020 10:30 |
|
cjs: I've been telling CTO for over a year now that it seems like it would be easier to just keep stateful systems running outside k8s and just use the k8s clusters for ephemeral containers that don't really need persistent volumes or anything between deployments. After another message bus outage (because CTO insisted on configuring the persistent volumes himself in openebs instead of paying the admittedly high prices for something like portworx) has him rethinking the whole approach, finally. I'm sure there's a way to do this, but our entire dev/ops/devops group is he and I, and I know less about k8s configuration than just about anybody, and he has about as much experience with k8s as I do, though he spends more of his work time actually working on ops issues, as I handle more of the dev side. ha, just got a slack message, I guess the k8s node died and was replaced, and somehow that blew up the message bus containers because the pvc's got rebuilt? I thought k8s apps were supposed to be tolerant of node failure, but apparently OpenEBS can't automatically reattach volumes to the new replacement node when it loses a node with attached volumes. or something
|
# ? Feb 20, 2020 18:05 |
|
Finster Dexter posted:cjs: I've been telling CTO for over a year now that it seems like it would be easier to just keep stateful systems running outside k8s and just use the k8s clusters for ephemeral containers that don't really need persistent volumes or anything between deployments. After another message bus outage (because CTO insisted on configuring the persistent volumes himself in openebs instead of paying the admittedly high prices for something like portworx) has him rethinking the whole approach, finally. I'm sure there's a way to do this, but our entire dev/ops/devops group is he and I, and I know less about k8s configuration than just about anybody, and he has about as much experience with k8s as I do, though he spends more of his work time actually working on ops issues, as I handle more of the dev side. Thread title, OP
|
# ? Feb 20, 2020 19:34 |
|
SAVE-LISP-AND-DIE posted:Thread title, OP lmao, CTO just dropped in slack #general "daughter fell down, gotta run" Last thing he said to me was "I need to login to the new node and install iSCSI and attach the new ebs volume." I do not have any AWS perms for accessing any of this. CTO took thread title advice I guess Meanwhile, production is still down
|
# ? Feb 20, 2020 20:03 |
|
|
# ? Apr 28, 2024 17:43 |
|
I need to write some user documentation for a project I'm working on. I was thinking about writing a bunch of markdown and then using pandoc to build/convert it all to pdf how terrible is this idea, is there a better way? the end result has to be pdf or word docs that can be distributed to the user
|
# ? Feb 20, 2020 20:09 |