|
Gazpacho posted:only managers believe that this is possible wrap the code in plugins, specify the build with inert data instead of mixing the two and hating life
|
#
?
Sep 24, 2017 03:57
|
|
|
|
| # ? Jun 5, 2024 02:34 |
|
|
i expect that the still-useful parts of mono that haven't been implemented in core yet will be migrated over with some effort and adaptation and mono will be phased out over another 5 years or something
|
|
#
?
Sep 24, 2017 03:59
|
|
|
pokeyman posted:wrap the code in plugins, specify the build with inert data
|
|
#
?
Sep 24, 2017 04:06
|
|
|
NihilCredo posted:for starters, if you add some packages to your project, then share that project with other people, when those people build the project they will always get the exact same package versions you used. this is not guaranteed with nuget  Sapozhnik posted:two features make a good package manager: thanks for the responses. i don't really have to deal with this sort of thing often because we rarely bring in external libraries, and check the artifacts into source control (!)
|
|
#
?
Sep 24, 2017 05:11
|
|
|
redleader posted:thanks for the responses. i don't really have to deal with this sort of thing often because we rarely bring in external libraries, and check the artifacts into source control (!) 
|
|
#
?
Sep 24, 2017 19:32
|
|
|
wrote a new blog post about something fun i discovered today http://blog.mecheye.net/2017/10/urg/
|
|
#
?
Oct 12, 2017 07:35
|
|
|
Suspicious Dish posted:wrote a new blog post about something fun i discovered today super interesting, thanks
|
|
#
?
Oct 12, 2017 13:29
|
|
|
redleader posted:
for all the poo poo this approach gets, there are still much worse things you could be doing.
|
|
#
?
Oct 12, 2017 13:35
|
|
|
Suspicious Dish posted:If you asked software engineers some of their “least hated” things, you’ll likely hear both UTF-8 and TCP. TCP, despite being 35 years old, is rock-solid, stable infrastructure that we take for granted today; it’s hard to sometimes realize that TCP was man-made, given how well it’s served us.
|
|
#
?
Oct 12, 2017 17:11
|
|
|
yeah, i sort of touched on it, but it's hilarious how many things either misinterpret "tcp packet" as "datagram" or try to rebuild datagram transport on top of tcp. tcp is actually the wrong idea, implemented super well. i don't know how well sctp works -- we don't have 40 years of using it in practice -- but i think most people want actual unlimited-size datagram transfer.
|
|
#
?
Oct 12, 2017 17:23
|
|
|
also that first paragraph was plagarized from tef http://programmingisterrible.com/post/69710734780/the-trouble-with-tcp-its-good-but-were-stuck
|
|
#
?
Oct 12, 2017 17:24
|
|
|
so the thing is that, while i generally agree about people actually wanting an unrestricted datagram protocol, there is so so much from tcp that you would need to reimplement
|
|
#
?
Oct 12, 2017 17:31
|
|
|
just use websockets op
|
|
#
?
Oct 12, 2017 17:55
|
|
|
rjmccall posted:so the thing is that, while i generally agree about people actually wanting an unrestricted datagram protocol, there is so so much from tcp that you would need to reimplement i said tcp is a bad idea well-executed. i don't care if you reuse / reimplement 90% of tcp's good execution, something that's datagram-based is still a better idea.
|
|
#
?
Oct 12, 2017 18:22
|
|
|
TCP is pretty drat good if only because it has prevented multiple generations of developers from writing custom packet deduplication, reordering, and retransmission mechanism in every other app around. I'll gladly take the possible slowdown of a few acks over a sliding window to 15 iterations of "reliable data transfer" libraries in JS for the browser. I find the tradeoff worthwhile even if you look at SCTP and you have to go "oh yeah that's nicer"
|
|
#
?
Oct 12, 2017 18:50
|
|
|
sctp is so good and it'll never be mainstream
|
|
#
?
Oct 12, 2017 19:45
|
|
|
hackbunny posted:sctp is so good and it'll never be mainstream backwards compatibility is a hell of a drug
|
|
#
?
Oct 12, 2017 19:58
|
|
|
i feel like for everything we use theres something better in some sense but not mainstream
|
|
#
?
Oct 12, 2017 19:58
|
|
|
Thermopyle posted:i feel like for everything we use theres something better in some sense but not mainstream see: opus vs. mp3; FLAC vs. wav
|
|
#
?
Oct 12, 2017 20:39
|
|
|
What's the better-but-not-mainstream alternative to PDFs?
|
|
#
?
Oct 12, 2017 20:39
|
|
|
VikingofRock posted:What's the better-but-not-mainstream alternative to PDFs? Electron
|
|
#
?
Oct 12, 2017 20:46
|
|
|
if you're scanning data, djvu is pretty light compared to PDF, but PDF does so much overall that making a less popular alternative has to fail at some aspects by virtue of not being able to make it do as many things.
|
|
#
?
Oct 12, 2017 20:49
|
|
|
rjmccall posted:so the thing is that, while i generally agree about people actually wanting an unrestricted datagram protocol, there is so so much from tcp that you would need to reimplement MononcQc posted:TCP is pretty drat good if only because it has prevented multiple generations of developers from writing custom packet deduplication, reordering, and retransmission mechanism in every other app around. I'll gladly take the possible slowdown of a few acks over a sliding window to 15 iterations of "reliable data transfer" libraries in JS for the browser. I find the tradeoff worthwhile even if you look at SCTP and you have to go "oh yeah that's nicer" i think QUIC still gives us some hope for the future, if nothing else than because google can and will solve the chicken and egg adoption problem by brute force. it's not usually presented as a datagram protocol but it supports multiplexed streams, and there is not really any difference between a multiplexed stream protocol with and an arbitrary-size datagram protocol. unfortunately the team responsible seems to be incapable of publishing it in a reusable form, but if it gets standardized that'll be moot soon enough.
|
|
#
?
Oct 12, 2017 22:24
|
|
|
Didn't Google still have issues with SCTP on researching it? TCP is an easy win as pretty much all $1+ NICs can accelerate it, UDP you are processing everything in software.
|
|
#
?
Oct 12, 2017 22:31
|
|
|
gonadic io posted:Electron im only finding some kinda javacript framework??
|
|
#
?
Oct 12, 2017 22:34
|
|
|
hackbunny posted:sctp is so good and it'll never be mainstream
|
|
#
?
Oct 12, 2017 22:36
|
|
|
Powaqoatse posted:im only finding some kinda javacript framework?? just implement a postscript to javascript transpiler. done.
|
|
#
?
Oct 12, 2017 22:50
|
|
|
lte and the like is built on SCTP so its not exactly unused
|
|
#
?
Oct 12, 2017 23:11
|
|
|
VikingofRock posted:What's the better-but-not-mainstream alternative to PDFs? joke answer: xps
|
|
#
?
Oct 12, 2017 23:18
|
|
|
the main problem with SCTP is that microsoft can't be bothered to implement it and consumer NATs don't understand it, so it can't realistically be deployed to networks outside your control. i've also seen suggestions that commercial networking gear handles it worse than UDP for some reason; no first hand experience though. if IPv6 adoption accelerates and microsoft decides to stop making things worse there might be change, but building something new on top of UDP, i.e. QUIC, is more realistic.
|
|
#
?
Oct 12, 2017 23:58
|
|
|
Suspicious Dish posted:i said tcp is a bad idea well-executed. i don't care if you reuse / reimplement 90% of tcp's good execution, something that's datagram-based is still a better idea. ironically, tcp is a bad idea reasonably executed well, mostly ok tcp as we know it went through many iterations tcp3 is where it split into tcp/ip which was a good idea at the time, but also ip fragmentaton exists so then there was tcp-reno et al being developed after congestion collapse then i guess all the syn-cookie problems, and the back and forward between stateless and stateful firewalls happened too and now bufferbloat and now we have a fossiled network of shitboxes at the last mile, great
|
|
#
?
Oct 13, 2017 00:03
|
|
|
quote:It’s incredible to think that 35 years of rock-solid protocol has had such an amazing mistake baked into it. it's incredible to think that a protocol wth 35 years of work is rock-solid
|
|
#
?
Oct 13, 2017 00:05
|
|
|
tcp is fine.
|
|
#
?
Oct 13, 2017 00:13
|
|
|
VikingofRock posted:What's the better-but-not-mainstream alternative to PDFs? HTML.
|
|
#
?
Oct 13, 2017 00:48
|
|
|
tef posted:and now bufferbloat  (from slides) but hey, plenty more problems anyway.
|
|
#
?
Oct 13, 2017 00:57
|
|
|
Ralith posted:if [...] microsoft decides to stop making things worse 
|
|
#
?
Oct 13, 2017 01:07
|
|
|
MononcQc posted:TCP is pretty drat good if only because it has prevented multiple generations of developers from writing custom packet deduplication, reordering, and retransmission mechanism in every other app around. I'll gladly take the possible slowdown of a few acks over a sliding window to 15 iterations of "reliable data transfer" libraries in JS for the browser. I find the tradeoff worthwhile even if you look at SCTP and you have to go "oh yeah that's nicer" i did work on game netcode for a bit once. making a game netcode begins by picking up udp and reimplementing all the poo poo you need from tcp, because tcp inevitably has that one extra thing in it you don't need for your game to run, and it prevents netcode from working smoothly. if you want to see what the world would look like in the alternate reality you describe, you only need to take a look into the hellscape of game netcode, there's 2-3 sane implementations that work decently well and then there's 3 billion variants of just really loving awful.
|
|
#
?
Oct 13, 2017 01:37
|
|
|
tef posted:it's incredible to think that a protocol wth 35 years of work is rock-solid the target audience of my post is hacker news posters and people on /r/programming. unlike anything those people have ever done, tcp does what it says on the tin.
|
|
#
?
Oct 13, 2017 01:39
|
|
|
what was that block/tower/stacking game that had people change the coalescing duration
|
|
#
?
Oct 13, 2017 01:40
|
|
|
|
| # ? Jun 5, 2024 02:34 |
|
|
Suspicious Dish posted:the target audience of my post is hacker news posters and people on /r/programming. unlike anything those people have ever done, tcp does what it says on the tin. fair
|
|
#
?
Oct 13, 2017 02:10
|
|