|
every time anyone from facebook talks about their ios app it sounds like the problem is actually too many of them working on it
|
# ? May 8, 2018 01:06 |
|
|
# ? May 31, 2024 09:25 |
|
Plorkyeran posted:every time anyone from facebook talks about their ios app it sounds like the problem is actually too many of them working on it yeah also they throw shitloads of money at very talented but very young programmers, who then quite predictably find very clever solutions to the wrong problems like they wrote a custom loader for their ios app long before they finished merging its component dependencies
|
# ? May 8, 2018 01:37 |
|
the facebook app must be written in some garbage ruby or javascript bullshit or something cause its got the worst performance for the simplest app.
|
# ? May 8, 2018 01:51 |
|
Shaggar posted:the facebook app must be written in some garbage ruby or javascript bullshit or something cause its got the worst performance for the simplest app. it's mostly in obj-c afaik. i think the performance issues stem from 99% of its functionality being in trying steal whatever data it can from your phone.
|
# ? May 8, 2018 02:24 |
|
although i guess technically obj-c does fall under the some ruby bullshit category
|
# ? May 8, 2018 02:33 |
|
Update on my WebAssembly adventures: for ridiculous reasons, WebAssembly.Memory, in Firefox, is backed by a giant virtual memory store somewhere that's "UINT32_MAX + INT32_MAX" large. This is passed through to the GC, who then goes "oh holy SHIIIIT WHAT i gotta GC NOW" and causes an emergency GC collect. This happens for every time the WebAssembly.Memory constructor is called.
|
# ? May 8, 2018 02:45 |
|
why is that being allocated on the gc heap instead of being pinned from an object there, jfc
|
# ? May 8, 2018 02:47 |
|
Because the GC needs an accurate count of how much space it has to know when it will run out? If there's something I want the GC to be immediately aware of it's a large allocation that could seriously impact its ability to allocate in the future. The memory itself is not part of the GC heap, it just informs the GC how much it allocated through a counter. https://bugzilla.mozilla.org/show_bug.cgi?id=1459761
|
# ? May 8, 2018 02:55 |
|
eschaton posted:yeah, I didn’t just talk with someone that it's pretty sad imo that you refuse to even acknowledge basic criticisms like "apple is kinda monopolistic sometimes" in terms of how they affect development for their platforms. then again you're a rich techbro so your societal privilege allows you to hide behind "but you aren't complaining about a benchmarkable thing in a menu i coded". still sad tho
|
# ? May 8, 2018 05:56 |
|
like just as a basic example of the hypocrisy here, it's great to be a proponent of accessibility systems. however this advocacy is kind of hilarious when you consider that most people with disabilities can't afford your fancy accessibility-enabled products
|
# ? May 8, 2018 06:02 |
|
Stringent posted:although i guess technically obj-c does fall under the some ruby bullshit category it obviously has some overhead, but obj-c is sort of shockingly not-slow for how dynamic it is. i've ported a decent chunk of obj-c to c++ (mostly for the sake of portability, not performance) and the difference between idiomatic obj-c and the c++ is less than a factor of two, and optimized(*) but not "why are you even still pretending this is obj-c" code is even closer [*] usually meaning adding __unsafe_unretained all over the place because refcounting is slow
|
# ? May 8, 2018 06:22 |
|
My nitpicky Mac development gripes (aka macOS is not Linux): * gdb does not work ootb for weird reasons. And after signing the binary (??) with a key I just made up (???) it still fails on most binaries for inexplicable reasons on trivial cases. * no strace and I can never remember the name of the macOS equivalent. * valgrind is touchy and reports (benign but noisy) errors in system libraries. * the default OpenCL comes with a completely useless CPU device that is anyway first in the list. Delete this. * the loving file system is case insensitive. I asked for a Mac mostly to try it out and get my software on Homebrew (most of my students use macs for some reason), and I've been fairly disappointed (also in general computery things like random crashes and peripherals not working). I've ended up doing a lot of my work on my home Fedora desktop over SSH. Did I miss the good years of macOS or something? This feels like a Linux.
|
# ? May 8, 2018 07:47 |
|
Athas posted:My nitpicky Mac development gripes (aka macOS is not Linux): alternatively: holding it wrong
|
# ? May 8, 2018 08:26 |
|
eh... fedora desktop is a mostly pain free experience. if you have to linux you could do worse.
|
# ? May 8, 2018 09:16 |
|
re linux: on the one hand it's currently at a state where I could install ubuntu 16.04 as a dual-boot on top of windows 10 on the other hand, it would still randomly cut and restore power to my usb sound device (causing loud static 5 minutes after I last played some sound) which I needed to do command-line tweaking to fix and when I wanted to uninstall the dual boot the official grub uninstall guides left me with an unbootable system, but i had a windows 10 recovery disk + easeus partition magic so that got unfucked without having to wipe my entire os disk so all in all, year of linux on the desktop! just not off the desktop, or with out-of-the-box working sound
|
# ? May 8, 2018 09:31 |
|
Athas posted:Did I miss the good years of macOS or something? This feels like a Linux. there was a period of time before touchbars you could play Doom on where macs had the latest and greatest hardware and arguably the smoothest *nix experience in a ui. at some point devs became more competent in linux, steve jobs died, and i loving hate my current MBP with its lovely keyboard and constant wifi/display issues. also its a great os if you like bash 3.2 and other outdated utils because apple is anal about GPL. but you can totally use brew and tap a keg to pour a pint into a tulip glass and harness the power of a lovely package manager that vomits emojis all over the term, if you want something from the last 12 years. more and more devs at my work are dumping off their osx laptops and asking for a dell with fedora, so maybe 2018 really is the year of linux on the des
|
# ? May 8, 2018 13:21 |
my 2017 dell has been by far the worst laptop ive had experience to use. my current favourite thing is that 5ghz wireless is not working, with any connection attempts hard crashing the wifi receiver
|
|
# ? May 8, 2018 14:03 |
|
you can't really complain though, that aspect of the wifi chipset is right there in the name (unless you actually have issues with one of the more normal variants, in which case i don't know what to tell you)
|
# ? May 8, 2018 14:40 |
|
dtrace is absurdly more powerful and useful than strace. it's one of the strengths of developing on macos, not a weakness.
|
# ? May 8, 2018 15:12 |
Cybernetic Vermin posted:you can't really complain though, that aspect of the wifi chipset is right there in the name (unless you actually have issues with one of the more normal variants, in which case i don't know what to tell you) no, i have the sku with idiot WiFi
|
|
# ? May 8, 2018 16:13 |
|
I wonder how much this will be Excel just copying Google Apps Script?quote:At the Build 2018 developer conference that's taking place these days in Seattle, USA, Microsoft announced support for custom JavaScript functions in Excel.
|
# ? May 8, 2018 17:37 |
|
Plorkyeran posted:dtrace is absurdly more powerful and useful than strace. it's one of the strengths of developing on macos, not a weakness. yes but they only know how to use strace so macos sucks for not having it because who wants to learn a new way to do things
|
# ? May 8, 2018 17:46 |
|
why would macos lack strace? solaris (where they got dtrace from) didn't get rid of strace
|
# ? May 8, 2018 18:33 |
|
Blotto Skorzany posted:why would macos lack strace? solaris (where they got dtrace from) didn't get rid of strace You mean truss, strace is some STREAMS bullshit from the early 90s there. All I know about dtrace is it doesn't seem to work with the output from my compiler (which creates statically linked binaries that don't link in anything, even libc, instead doing raw syscalls) buuuut my setup is old so I might not even have 64 bit dtrace I guess? and also obviously this is a hyper niche case. Similarly the Linux subsystem for Windows won't run the Linux output.
|
# ? May 8, 2018 18:42 |
|
dtrace definitely can log raw syscalls on macos
|
# ? May 8, 2018 18:44 |
|
Plorkyeran posted:dtrace definitely can log raw syscalls on macos In general, yes, but it seems to want to hook or attach to the process in some way first and instead, obscure error message I don't recall off the top of my head.
|
# ? May 8, 2018 19:18 |
|
as someone that jumped from ant to gradle, and thinks gradle is therefore wonderful, what's the argument that maven is better than gradle?
|
# ? May 8, 2018 19:28 |
|
maven is declarative instead of procedural
|
# ? May 8, 2018 19:31 |
|
shaggar is right
|
# ? May 8, 2018 19:37 |
|
Peeny Cheez posted:shaggar is right
|
# ? May 8, 2018 19:47 |
|
Helicity posted:as someone that jumped from ant to gradle, and thinks gradle is therefore wonderful, what's the argument that maven is better than gradle?
|
# ? May 8, 2018 19:58 |
|
maven: the choice of discriminating devs who check databases into source control
|
# ? May 8, 2018 20:00 |
|
Gazpacho posted:well the argument that keeps being thrown at me itt is that maven provides a build strategy to those who don't really care how their poo poo gets built. the moment you do care, maven will turn on you if your code follows standard patterns, either maven or gradle will just build it and you don't have to think about how that happens. in this situation it really doesn't matter much which you use! if your code doesn't follow standard patterns, maven will encourage you to correct this so you end up with maintainable code, while gradle will encourage you to copy-paste 25 lines of code you don't understand in a language nobody knows into your build script, and pray nobody ever tries to build your poo poo with a slightly different version of gradle because god help you if they ever change part of the api (spoilers they change random bits of the api in non-backwards-compatible ways all the time) which is why anyone sane wants other people to choose maven
|
# ? May 8, 2018 20:31 |
|
you don't understand my build is special
|
# ? May 8, 2018 20:34 |
|
the idea of a universal "standard" build process is defectiveleftist heap posted:beep boop the apache foundation is always right
|
# ? May 8, 2018 20:37 |
|
leftist heap posted:you don't understand my build is special
|
# ? May 8, 2018 20:43 |
|
as far as maven is concerned any build that produces anything other than a single library jar is "special". so... ... yes?
|
# ? May 8, 2018 20:49 |
|
Gazpacho posted:well the argument that keeps being thrown at me itt is that maven provides a build strategy to those who don't really care how their poo poo gets built. the moment you do care, maven will turn on you maven is for if you care if it gets done correctly every time. if you want to build it differently every time use gradle or ant.
|
# ? May 8, 2018 20:51 |
|
Gazpacho posted:as far as maven is concerned any build that produces anything other than a single library jar is "special". so... ... yes? no
|
# ? May 8, 2018 20:52 |
|
|
# ? May 31, 2024 09:25 |
|
Gazpacho posted:as far as maven is concerned any build that produces anything other than a single library jar is "special". so... ... yes? you can use all sorts of maven plugins that produce secondary artifacts tho???????????????????????????????????????????????
|
# ? May 8, 2018 20:54 |