wincucks btfo
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
pic related
>The system that builds windows is a bunch of 30 year old perl scripts from the 1990s. No one knows what they do. They have tried to replace this build system 4 times. The most recent attempt happened while I was working there and was called "Apex". It worked great, solved all the problems, and was totally fantastic, except that it couldn't compile windows. No one could figure out how to get windows to compile on it. This is because no one knows how to compile windows. They eventually shelved Apex because no one could decipher what the flying fuck was going on with the perl scripts.
>They still use the perl scripts.
this is why I started running OS/2 five years ago
I am unironically intrigued and I want to hear more
What the fuck is it with large companies and legacy Perl code.
obviously explorer is a retarded COM+ nightmare, still completely incompatible with C# after 15 years
>>53741805
[citation needed]
>>53741805
I'm blown away. What is the benefit of using Perl to compile something? It's sounds insane.
>>53742057
>I am new u cannot prove shit
>>53741907
>>53742074
>i read it on the internet, so it must be true
>>53742105
>it was the windows 10 control panel
Here I thought MS actually rewrote and updated the OS from scratch with each major iteration, instead of running on code from windows 95. What the fuck MS.
underrated thread
>>53742146
I was under the same assumption that each Windows was written from ground up starting with Windows XP. Only to find out Microsoft probably has everything running in a modded DOS system that emulates a second O/S
I'm intrigued to know more about this shit
>>53742185
>Catalog
how can anyone be this stupid
that's like saying the US government completely re-constructed itself from scratch every presidential election
>rewriting millions of LOC every time
>>53742207
and I honestly have no idea how ">Catalog" even got there
[8:00:33 PM] <Person> DO YOU PEOPLE WRITE SQL LIKE "SELECT * FROM `TABLE`" THEN FILTER CLIENT SIDE?
[8:00:33 PM] <Anon> not with scaling it to actual real world data
[8:00:40 PM] <Anon> actually they do sometimes
[8:00:45 PM] <Anon> like AT MICROSOFT
[8:00:51 PM] <Anon> where we kept having PERFORMANCE ISSUES
[8:00:56 PM] <Anon> BECAUSE IDIOTS CAN'T FUCKING DATABASE PROPERLY
[8:01:07 PM] <Anon> i mean really I question a lot of the hires over there
[8:01:19 PM] <Anon> then again one of them sent me example code as a screenshot embedded in a word doc so perhaps my expectations are too high
>>53742207
it's not like that at all because anyone can write this stuff if they have enough time, and windows clearly has the time in man hours
it must mostly have to do with backwards compatibility which is literally the reason why pretty much all technology is ugly under the hood
[7:02:44 PM] <Anon> i will never trust microsoft on anything performance related ever again after seeing their leadership's priorities
[7:53:50 PM] <Anon> that reminds me of that time where i re-factored our spaghettified codebase at microsoft into a much cleaner, organized whole
[7:54:09 PM] <Anon> I was then reprimanded by my manager because "we don't just refactor code unless it benefits the customer"
[9:49:14 PM] <Anon> different teams under the same leadership will have violently different opinions
[9:49:29 PM] <Anon> which can and sometimes do devolve into actual shouting matches
[9:49:51 PM] <Anon> which my lead loved to recount to me when he saw them but I never got to see any myself (because I hated meetings and usually got out of them)
[9:51:24 PM] <Anon> so team A wants to do X and team B want to do Y
[9:51:31 PM] <Anon> and 99% of the time this is resolved by doing both X and Y at the same time
[9:51:34 PM] <Anon> consequences be damned
[9:53:08 PM] <Anon> so when microsoft is doing all these contradictory things
[9:53:16 PM] <Anon> it's because no one has any semblence of a "vision" there
[9:53:29 PM] <Anon> it's all super vague "reinvent productivity!" stuff
[9:53:54 PM] <Anon> like if you ask any one microsoft person what the vision of the company is they'd probably either quote satya being vague or be like "Make cool stuff for people???"
[9:55:05 PM] <Anon> it's way too big, no one knows what's going on, no one agrees on what to do, everyone invents stupid things, they haev so much money they don't know what to do with it, and they never fix their code, prefering instead to just make some new product that supplants the original because no one can figure out how the code even works
[9:55:57 PM] <Anon> they try really hard to hire smart people
[9:56:05 PM] <Anon> who sometimes succeed in making good decisions
[9:56:37 PM] <Anon> but are ultimately wasted by idiot managers who insist on doing incredibly dumb shit
So that's why I find old as fuck UI in between the new stuff.
>perl build system
Can't they, you know, load it in a debugger and go through it step by step?
I'm sure giving the task to some intern to document would take just a couple of weeks.
Post more snippets, this is great stuff.
Raging devs produce such juicy reads, at some point somebody needs to dig into the ffmpeg/libav debacle,
Seriously what will happen when old devs that know how to handle the spaghetti code die?
What will MS do when they no longer can maintain such intrincate projects.
[4:03:19 PM] <Anon> i already told you
[4:03:23 PM] <Anon> at microsoft, you don't fix bugs
[4:03:26 PM] <Anon> it's a waste of time
[4:03:32 PM] <Anon> you need to implement new features to keep the managers happy
[4:03:47 PM] <Anon> unless one of the managers hits a bug and it's reeeeally bad, then maybe it'll get classified as P1 and someone MIGHT fix it
[4:03:49 PM] <Anon> maybe.
microsofts face when
I feel like oliver twist asking for more.
More, please.
>>53742105
>>53742137
Fucking glorious
>>53742277
is this the same guy that was posting on reddit a couple of months ago, he made the same argument that at MS they have zero motivation to refactor or improve performance
I'm trying to get more snips by skype keeps crashing when I try to load old messages.
:^)
>>53742185
The DOS-based Windows line died out after Windows ME. 2000, XP, and everything since have all been based on the NT kernel, which was never based on DOS.
>>53742269
>backwards compatibility
This. A significant amount of Windows is purely for compatibility. I think they cut out a sizeable chunk of it in 8/8.1 because no one used it.
>>53742381
>I think they cut out a sizeable chunk of it in 8/8.1 because no one used it.
No way in hell. For 20 years the entire system has been built around backwards compatibility above all els.
I lost my shit at [spoiler]____-1______[/spoiler]
>>53742367
You are a glorious human being sitting on a potential gold mine of entertainment.
I salute you.
>>53742434
Agreed. Plz dump.
The samefag is strong with this thread.
>>53742483
(You)
>>53742452
>>53742434
Table-scanning my sqlite db for more gold.
>>53741995
Nobody can read them, so they stay.
>>53742068
You mean in the first place?
It's still better than no build tool or UNIX (TM) GANOO makefiles.
> does it build?
> who cares I'm rich bitch
mfw that sounds exactly like my job at autodesk
>>53742518
>hating on makfiles
They're still the most sane C/C++ build system.
>>53742588
makefiles get a bad rep because people confuse them with gnu autoconf/automake which actually are insane
>>53742575
I am so sorry for you, anon. Fucking autodesk and solid works. Jesus.
Still has images in the file picker menu though. MS wins again.
>>53742575
>autodesk
Ayy lmao
Autocad is one of the most wtf, programs I have every used. Click something, curser is gone, change tabs, curser is back again.
sqlite> SELECT body_xml FROM messages WHERE author="**********";
ah yes, microsoft at its finest
i'm pretty sure skype does everything in every way imaginable and then invents several unnecessary ways of doing it too
there are plenty of cases where microsoft has pushed REALLY BAD updates
like that one update where they broke their SQL server for 3 days
It's usually more applicable to businesses
there will be no windows 11
speaking as a former microsoft employee, they are turning windows into a service
I should know, I helped build the system that lets them do that
Windows 10 is the last proper release of the operating system. From now on they'll just update the OS using the swap system
there were also rumors of it eventually being completely free
This is why they kept experimenting with putting ads into the start menu
Eventually they will rebrand Windows 10 as just Windows
they're tired of competing with themselves, so now everyone will get all the updates and they'll just constantly put out new features
they still support XP because of the government
They haven't decided how any of that works
I got in trouble while at microsoft because I spent two weeks refactoring the code
fun fact: Microsoft spent 3 years trying to rewrite the entirety of windows on top of a managed kernel. This is why they took so long to ship vista. They spent an additional 10 years still attempting to get their managed kernel to work before finally giving up last year.
Vista's codebase got so fucked up, they reverted the entire source tree to XP and then rebuilt the entire OS on top of it
They literally spent 3 years trying to do something insanely dumb and then threw it all away
I really like how half of MSDN is down right now while i'm trying to look up a windows system call
>>53742653
fyi it also doesn't make any profit and is a huge money sink (it's been like that for at least coupe of years now)
>>53742284
Lurk in #mpv and #mpv-devel if you want plenty of developer rage
>>53742588
Makefiles are great for small to medium sized projects with not much configuration. For more complicated stuff, I like Ninja better. It's generally similar to Make but gets rid of a lot of the annoyances, and it's by far the least shitty of the build systems I've tried (cmake, scons, shake, and I think one more that I've forgotten).
>>53742367
Linux or Windows client? I can load hundreds of thousands of messages on the Linux client, going back years.
>>53742764
Linux. This chat sees about 1000 new messages per day.
>>53742749
Don't cmake and scons just generate makefiles?
I've used way too much IIS
because... microsoft
you need to make sure you have all the proper things installed
so like, the .net framework 4.0 or 4.5 or whatever, the HTML components, the ASP.net components, etc.
there's a LOT of them
I don't actually know what half of them do, we just had a list while I was at microsoft that said "install all these things to make ASP.net magically work"
(we had a lot of lists like that)
Not if it's running windows 10
give them a few more months to not fuck everything up with their updates
ATM windows 10 is a lottery
you'll probably be fine
but there's a 1% chance with each update your entire computer will be unusable
they fired all their testers
while I was there, I might add
Also the software devs are expected to test things
(hint: they don't)
we were once told to use something because they'd seen it work "on a dev machine once"
like, that's the standard
"does it probably work? SHIP IT"
there's a reason I quit that job
microsoft has no QA anymore
the insiders are the QA
they just throw shit at customers dumb enough to test things for them
well that's how it was at the redmond campus, at least
all the testers, all gone
they completely removed the role of "software developer in testing" and now everyone is just "Software engineer"
there literally isn't a "tester" job description there anymore
Unfortunately despite working at microsoft, I still don't know windows 10 permissions well enough to help you
other than to say: god I'm sorry you're using windows 10
there's a reason i'm staying on windows 7 until the end of time
I can gaurantee you no one at microsoft knows either
because thats how we fixed things
"Well we don't know what went wrong, just try reinstalling/rebooting!"
hahahaha
you're kidding right
you have no idea the crazy shit that made it into the final build
I HELPED WRITE THE CODE
I HAVE SEEN THE ABYSS
>>53741805
Wow, that's amazing!
> buy mac
> everything just werks
Thanks Apple.
>>53742803
they could've done a lot of things, but you have to understand, at microsoft, you aren't ALLOWED to fix bugs unless they are high priority
if you go and fix some random bug that wasn't assigned to you your manager will have a "discussion" about priorities with you
even if it only took you like 2 hours
i gaurantee you some fucking idiot already knew this bug existed
and it was labeled P2
and no one fucking fixed it because they had higher priorities
like implementing fucking stupid features no one will ever fucking use
because that's what the managers want
I CAN GO ON ABOUT THIS ALL DAY KNIGHTY
Setting it as P1 would make the manager's manager's manager's manager mad because it clearly isn't a data-loss bug or causes any kind of catastrophic behavior, it's just "kind of annoying" so obviously it's P2, but no one fucking fixes P2 bugs because they always have something more important to do
We once had a bug where all your drives stopped showing up and we still shipped it to all our devs inside microsoft on the slow ring and pissed them all off
and when they complained, our boss's boss's boss's boss's boss told us to shut the fuck up and deal with it
press F5
file a bug with microsoft
it will get accepted, put on P3 because it was filed from outside and never fixed fucking ever
windows 10 has serious driver issues
because 3rd parties haven't fixed their shit yet
which is to be expected
because the exact same thing happens literally every time they increment the kernel version number
it's the same root cause, actually
vista incremented the kernel version from 4.x to 5
this broke every single driver that looks for 4.x
even if it would have worked on vista
windows 10 incremented the kernel to 10.x
soooo
same thing happens
it's even worse now because it's 2 digits
which breaks even MORE things
when i heard about that my whole team was laughing about how it was going to be a disaster
desu still works better than any linux distro I've used
The only linux I never had any problems with was Debian but it was server w/o desktop environment do I guess it doesn't count
>>53742787
Cmake does, scons doesn't. But the actual process of setting up the build system is way different between cmake and make.
I must have missed it, where are these juicy tidbits coming from?
I'm enjoying this thread
I want to read more
>>53742851
oh god metro
let me explain metro
metro uses flat design as an excuse to be lazy, because no one at microsoft knows how to design UI
No one WANTS to design UI, because designing UI is hard
so they use flat design as an excuse to make everything boxes so the program managers can make shit in powerpoint and give it to the programmers
i'm not making that up either, they literally design all the windows UI in powerpoint
they gave us powerpoint slides and were like "make this"
that's why it's flat.
It's powerpoint.
it's complete fucking shit
windows 10 was built for microsoft employees
and microsoft employees are fucking idiots like you'd never believe
we found what was apparently test data submitted to one of our production databases
We knew it was test data because the reported architecture was "poop"
i always have to use google to search MSDN
because Bing can't search it's own goddamn documentation database
this became relevent while i was employed at microsoft
because y'know, have to use all the microsoft stuff
and our OWN SEARCH ENGINE couldn't search OUR OWN DOCUMENTATION DATABASE
it was infuriating
i had constant problems
it probably didn't help that half the time they'd be like "oh you're an employee HERE USE THIS BROKEN BETA VERSION"
'WHETHER YOU WANT TO OR NOT!'
i hate dogfooding
it didn't fix anything faster
it's where you force all your employees to use your untested software
e.g. windows 10 was built on top of windows 8
while we were using it
I got to watch them slowly try and take windows 8's UI and violently stuff it inside windows
I don't think I can describe that specific scenario as rape because my entire employment experience was
>>53742735
Speaking of skype...
http://www.secdev.org/conf/skype_BHEU06.handout.pdf
>>53742851
>KNIGHTY
...not the fimfiction.net dev Knighty, right? Surely not?
>>53742936
so uh
why did they have to redesign if literally no one could do it
>>53742588
>>53742627
>>53742749
That's what faggots believe, but makefiles are technically cancer and easily rivals Perl and XML in readability.
There is no such thing as trivial and functional applications, therefore there will be a point at which your build system should have been a cross-platform distributed library for a real programming language to begin with.
>>53742787
Yes and so are GNU autotools, and they are hepatitis C, therefore.
Meta build systems are a thing that shouldn't exist.
And people repeat this mistake over and over.
It's like with package managers where people hate programming language package managers - and rightfully - for being crap but at the same time think that bistro packages are a good solution and neglect the fact that even most professional bistro maintainers are not up to the task and that ecosystem subnodes would be better.
I'm reading all this, right now, on a netbook running Windows 10. /popcorn
>>53742919
OP's posting logs of a skype chat with a disgruntled ex-Windows dev
>>53742959
A simple elegant BSD-style makefile isn't that bad.
>>53742936
>architecture "poop"
next level of kek
>>53742952
[spoiler]is that you vic?[/spoiler]
>>53742839
OS X, the NeXTSTEP skin shoehorned to run on HFS+? I really want to read some horror stories about that.
If I had to bet, I'd say that the guy if ever worked at MS didn't quit but was fired and that's why he rants about them
>>53742959
>makefiles are technically cancer and easily rivals Perl and XML in readability
Kek, what? Makefiles are dead simple if you just need it to build a pile of .o's and an executable or two.
It's only once you start trying to do complicated shit that it really goes to hell.
For instance: https://github.com/rust-lang/rust/blob/master/mk/install.mk
That's a randomly chosen file from Rust's ~4500 lines of makefile-based build system.
>>53742969
At some point it will be, though.
And I can't agree on that one, anon.
Judging from https://www.freebsd.org/doc/en_US.ISO8859-1/books/developers-handbook/tools-make.html
If something simple declared needs cryptic shorts, dollar signs and other crap and indentations it shows miserable language design.
I'd rather write Perl, and I'm not a fan.
>>53743036
t. Pajeet
>>53743040
I thought the rust compiler was written in rust and used cargo. What the fuck is that?
>>53742936
Microsoft knows that being a giant OS monopoly is not going to help it's bottom line anymore
The anti-trust case against them also scared them shitless and changed how a lot of things worked
Originally the idea was to get everyone on to windows, but now the idea is more to focus on locking certain technologies to windows but giving windows out basically for free
then, because windows is no longer the sole profit driver, you can make even more money by porting all your other applications to other operating systems
microsoft devs themselves are usually okay
(except windows update and some other teams)
but it's hard to seperate stupid programming from stupid management decisions, and there are SO MANY BAD MANAGEMENT DECISIONS
but yeah all of microsoft's high level management is a truly impressive brand of fucking stupid
the increasing amount of clashing i had with the management is one reason I left
i also got really sick of lying every single day about things like "do you have any concerns about the company"
>>53741805
I can confirm. I'm not in Windows, but the image is accurate. c++, xaml, resourceId, and hooks is exactly like our process (we use that all in office. Even in the new stuff). Some programmer thought of this retarded intricate system that will "totally save time" but has turned into a weight that slows everything down.
It's so deep that the only way to get rid of it is to start from scratch, but we'd lose backwards compatibility. We can never lose backwards compatibility... Especially in office
>>53743010
No idea who that is, I just recognized the name "Knighty" because of how much time I waste on the /mlp/ fanfiction general.
Are you really telling me THAT Knighty just happens to be on the same chat with this MS guy?
>>53743081
No, cargo is mostly a package manager with minimal build functionality. That's what they get for not building a better build system.
>>53743040
They are still shit compared to what they could be.
>Spergs in charge of build systems
Wait which control panel are we talking about?
Aren't there like... 2 in win10?
>>53743087
In PM with sauce:
>So, one of the other groups we worked with had their own problems. They were in charge of keeping track of which users were registered to which computer so we'd know who to send out new windows versions to. They did this with the help of another group that gathered up all the user information and sent it to them. Unfortunately, the other group decided that their API would only work on 100 users at a time. This is the same API that was going to be called when 14 million people got windows 10 on release. So, they had to build a giant cache of all the user results before they could actually send them to the other group to be stored, basically re-implementing half the functionality, while some poller ran in the background, agonizingly calling this API with 100 users tens of thousands of times whenever a big update went out.
>>53743103
Fucking hell. Isn't there a single systems programming language with a sane build system?
>>53743112
>While building the database that tracked windows builds, windows update thought it would be a great idea to make an update database keyed to the ID of our new kernel update method... even though they'd still have to serve updates to old systems that could not use our new system and therefore did not have the new ID format. So instead of actually changing their database to key on something that ACTUALLY EXISTED, we had to build an API for them to generate fake IDs until they got time to fix their database structure. Two months before the release of windows 10, they were still using it, and had generated over 200000 fake IDs.
>Or that other time, when we managed to take down the entire windows update system, including the windows store for windows 8, because we published too many new builds to the system. They had a poller running on the build metadata that did not check to see if it was already running, so too many builds got published, it'd take longer than whatever the polling interval was, and it'd just start an endless cycle of spawning new pollers that began tripping over themselves until the entire system crashed (if you recall a time back in early 2015 where the store stopped working over the weekend, this is why).
>As far as I know, they never fixed the poller bug, we were just really careful not to send too many builds.
>>53743081
rustc has been built with makefiles since long before cargo was created. Plus cargo, like all language-specific build systems, is absolute shit if you're trying to do anything remotely complicated.
>>53743028
I'm sure the code is still in a better state than NT's, though.
Apple seems to actually bother throwing out backwards compatibility. It's seen as a jew move by many (and it probably is for Apple), but it keps the code cleaner.
>>53742952
Yes.
>>53743104
Yes, along with the 20 different context menus
>>53743120
No, because they never learn.
>>53743126
I always wondered how windows update could manage to be 50 times slower than apt
part of me wishes I didn't know
>>53742277
this is pretty sad, really. every windows release i wonder why they keep adding useless features instead of fixing their shit, but this is why
this is why windows 10 has a 'personal assistant' but still only supports filesystems from the early fucking nineties and before
>>53743095
Even in the UWP/iOS/OSX/Android Office?
>>53743193
Come on man, apt isn't that slow.
>>53743193
I never understood this. I kinda hacked together a script to run windows updates on an offline windows install just to do updates. 150 updates in 30 minutes. Fuck windows update.
>>53743168
>>53743126
What the fuck , I'm beyond words.
Its like MS did so many abstractions that its literally biting them in the ass due to nobody knowing how they really work
Me coming from netsec side of things , how do they do the privelage escalation and other exploit paching ? Is there any data on that ?
>>53743202
Okay, that I don't know. Those guys are over in India or Vancouver. I'm just talking about desktop and universal
>>53743225
>microsoft
>patching exploits
>>53743158
It's obviously bound to be better, but there's a 0% chance there isn't some fucking insanity going on in there, especially in the early days when stuff like Carbon was still relevant. Finder was Carbon IIRC until 2009.
>>53743242
>implying that they don't pach the exploits that give you NT/SYSTEM instantly
>implying that they didint fix most dll hijacking vulns that allowed you to bypass UAC and give you escalated privelages
They seem to roll out those types of updates quite quickly
>a ragtag group of virgins led by a 300 pound toe cheese eating autist managed to create a better OS than a highly trained team of paid professionals
>>53743296
>GNU code
>good
lmao, keep in mind GNU has a version of true.c that has 60 something lines
I have to sleep or I'll pass out and this thread will probably fall into the archive before I wake up but thanks OP, I enjoyed this and look forward to waking up to a nice fat thread of juicy dev rage.
>>53743028
>>53743158
Here's the thing though, Apple actually cares about their developers. Apple's developer tools are really nice, and their APIs are all really nice and sanely set up. Cocoa is beautiful compared to COM+
so this thread is from a street shitter that got canned or what?
>>53743276
I'm sure they have a couple of 'god' engineers whose word is law, and who can do things no matter what the managers say.
There's only so much they can do, though, time is short.
>>53743340
nah, he got off the wild ride by himself
>>53743339
>swift
>xcode
I hope you're joking
>>53743322
And every single line of that is documented.
>>53743362
Not in the slightest. Care to elaborate?
>>53743376
Yeah, and it's still completely unnecessary.
More bloat = more vulns.
true.c should be what, 2-3 lines? Literally every other variant out in the wild is like this, but GNU had to go all special snowflake with it.
I don't even want to think about the emacs codebase, it must be a complete fucking mess.
>>53743239
>universal
Aww, I was hoping the universal Office apps would ditch all the legacy garbage. This is upsetting.
Some things said ITT are true (MS is huge and teams fighting within it) but the first pic in OPs post is a fucking lie.
The build system for 10 is new, and without it there would be no Win10 insider stuff. And the settings app is an UWP, and writing a combo box for it is rather simple, with no magic attached.
Y'all been rused.
>>53742277
>[7:54:09 PM] <Anon> I was then reprimanded by my manager because "we don't just refactor code unless it benefits the customer"
Bullshit. Check ins are gated. If the refactor wasn't authorized the code check in and codelook review would have been rejected.
Not that Microsoft doesn't have issues, (and Windows 10 it's own set), but someone is blowing smoke...
Source - I work under one of the OSG build masters.
>>53743377
Not the above poster but Xcode is absolutely bloated and buggy as fuck. Random freezing, random errors that are only fixed by restarting the program. The UI is complete trash too. They have to rework the entire program it's not usable.
Swift on the other hand is pretty cool, not sure why the >maymay arrows.
>>53743394
I just looked at the code for it and the complexity for that comes from --version and --help flags, and the fact that it's actually also 99% of false.
Here's the code for false.c in its entirety:#define EXIT_STATUS EXIT_FAILURE
#include "true.c"
>>53743398
My feelings exactly when I was hired! I'm not very high on the totem pole, but I'm tempted to see if I can refactor it all in my spare time (just one of the universal office apps) in C# and present it to my boss. I know they would _never_ accept it, but it would be nice to prove that a clean rebuild is possible. And my doing so I would have deep knowledge of the insane system and would be able to work better with it.
>>53743413
If it's anything like office it's not. Xaml is the dethronable king.
>>53743492
>Let's put the functionality of a program called false in a program called true and add even more flags to our programs!
Brilliant design by the amazing GNU team.
And no, I'm not saying true.c or false.c will have exploits because of this, I'm saying this is representative of the GNU design philosophy.
>>53742068
My company uses a python build system that then generates makefiles. It is supposed to generate build dependencies so you can build the system in parallel.
It sucks, especially when you purchase an API that already has a makefile. You must then turn that makefile into python that hopefully generates a makefile similar to the one you wanted (and is somehow shoved into our monolithic binary)
>>53743459
This.
Also
https://www.quora.com/Does-anyone-feel-that-the-software-quality-of-Apple-is-going-down-the-drain-Why-is-this-happening
>>53743418
How close together are the Desktop and Mobile teams? Is it true that it was all merged or is that just PR talk? W10Mobile is still based on Windows Phone 8 code IIRC?
>>53743492
>>53743540
Also forgot to mention, WHY does --version and --help even exist in most GNU programs when there's man pages and gnu's very own info (lol)?
No wonder Android doesn't use it.
What am I looking at here? KEK
>>53743557
--help is a lot more compact than the full manpage, which is great if you mostly know what you want but don't remember the exact flag name.
>>53742367
>using the smiley with a carat nose
>>53743540
I think the most questionable part there is that it doesn't actually return a 0/1, but an EXIT_SUCCESS/EXIT_FAILURE. It's probably defined in sys/types.h, but I haven't looked at it, and I can safely bet that it's something with 40 years of unnecessary backwards compatibility to it.
>>53743575
Because typing man <command> and / and typing the flag is so hard. A well written manpage also has all the flags in the same place, and they're well-written.
It's really not that hard.
>>53743547
They're separate teams, but work from a common set of API's and guidelines. While the goal is unification, each platform does present a unique set of challenges, and there are already SME's on each team. It doesn't make sense yet to combine everything. But there are methods for knowledge transfer, and much like the platform, the teams will be unified eventually as well.
>>53743541
>purchase an API that already has a makefile
...what in the fuck? I assume you mean "purchase a library"? But then who's selling libraries that come in source code form?
But as for the rest of it, I agree, that's pretty dumb. If you have a third party package, you should just invoke its own build system from yours, not go all Dr. Frankenstein and try to merge the two together.
>>53743594
>but don't remember the exact flag name
>>53743610
Yes, and man pages usually have an entire section dedicated to flags.
You can usually search for "option" and find them instantly.
>>53743557
fwiw, busybox's true.c with comments omitted:#include "libbb.h"
int true_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int true_main(int argc UNUSED_PARAM, char **argv UNUSED_PARAM)
{
return EXIT_SUCCESS;
}
>>53743601
The company only buys/licenses source code that it can compile itself. I usually think of a library as a shared obect or statically linked library with headers. But yea, it causes a headache.
>>53743655
That's way better.
BSD's have an even saner version. So does Plan 9 probably.
Don't know about Toybox.
wait, even microsoft has no fucking idea how windows works?
does this make it more or less cool?
>>53743672#! /bin/sh
# $OpenBSD: true.sh,v 1.2 1996/06/26 05:41:53 deraadt Exp $
exit 0#! /bin/sh
# $OpenBSD: false.sh,v 1.2 1996/06/26 05:32:50 deraadt Exp $
exit 1
>>53743709
That's old.
http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.bin/true/true.c?rev=1.1&content-type=text/x-cvsweb-markup
>>53743629
Not sure how you're still not getting it.
>OK, I need the thing to display the stuff
>I know there's a flag for this, --show or --verbose or something
>man thing
>search for --show
>no results
>search for --verbose
>no results
>start scrolling through 20 pages of detailed flag descriptions
>eventually find --print
>thing --help
>it outputs an alphabetical listing of every flag that fits onto one screen
>quickly notice there's no --show or --verbose, but there is a --print, that's probably the one I want
>>53743721
You can start by being smart and searching for long options first.
It seriously is not that hard.
>>53743703
Windows was at first not a Microsoft OS.
IBM requested an OS for their home PC and Microsoft bought the OS from another company.
>>53743721
>>53743730
Also you do know you can press N to go to the next result in most pagers, right?
>>53743655
no --help, --version or --license (suspiciously missing in the gnu version too). you needs don't know anything about writing user friendly software
>>53743766
man true?
>>53743730
>>53743764
Are you literally retarded? The entire premise here is that you know the functionality you need but don't remember the name of the option. In which case you don't need the full details of the manpage, and it's better to have a quick summary of the flags to remind you of the syntax, rather than spread them over 20 pages of text with every minute detail of every feature.
>>53742289
pic related
Bump to keep it going
>>53741805
>>53742105
>>53742255
>>53742277
>>53742311
not sure if legit or just advanced shitposting. is there any source besides IRC logs?
>>53744166
who in gods name would want to attach their name to this kind of dirt
What did people expect when you have high level management that doesn't even know how to write a single piece of code? the larger a company is, the larger the distance between coders and management. Sadly this is the average nature of current day large tech companies, and it completely destroys the quality of work being put out by these companies.
>>53743459
Ok you have a point with xcode but actual APIs and stuff like Swift is pretty smooth and cool for developers
>>53743655
What exactly would this be used for? It just returns exit success...
>>53744867
It returns exit success.
>>53744867
Shell scripting and shit.
Working for Microsoft as a programmer must be as horrifying as Dante stepping into Hell.
>>53742114
>winshits will defend anything by any means possible. Even in the face of undeniable truths.
>>53744867
Same reason a handful of programming languages have an identity function in the standard library and Python has "pass", sometimes you need to fill some sort of placeholder with something that doesn't actually perform any operations but the spot it occupies is not allowed to be left empty
>>53744867
it's just people getting autistic about an optimization to already slow as shit shell scripts for a program that isn't even run because it's builtin to every modern shell
Is it odd I sometimes fantasize of a windows built largely from wine and mono? A bunch of unpaid neckbeards managed to make a system that substitutes an impressive amount of largely undocumented, closed source, and works fairly well. If they could work with the source, instead of reverse engineering it, could we actually have a version of windows worth a damn?
>>53745335
Reactos?
>>53744867
I guess that if the program is terminated correctly (AKA nothing forcibly killed it) it has the time to execute this, so if someone wants to make a script with it they can use
>If ERRORLEVEL=0
To do something only if the program finished correctly. On the other hand, if the process can't execute this, it aws finished forcibly, posibly due to an error, and the script that depends on the output of the program can execute it again until the process isn't finished.
So, is someone taking screenshots?
i'm sure some linux-fag want to post in a future windows appreciation thread.
>>53745352
Very slow progress, still reverse engineering, and aiming for compatibility with Server 2003.
>some code monkeys fucked over literally FUCKING BILLIONS of people around the world by obfuscating their code without any documentation, probably so they could keep their jobs secure by necessity, because of hostile backstabbing office politics
>literally
>billions
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>nonfree software
>>53745258
What undeniable truth? It's literally a screenshot from god knows where/IRC. Come back when you have a credible source.
Explain to a non-programmer what is happening ITT
I want to have fun too
>>53745503
Don't be dumb, dude. Read the thread.
>>53745503
The method to modify and build Windows from source is overly complicated and problematic.
Not a huge surprise, it's a huge paste-together of tons of legacy and glue code. Actually impressed that they can get the fucking mountain of shit working as well as they do, honestly.
>>53741805
For some reason, I really REALLY want to see those perl scripts.
I want to peer into the abyss.
>>53745815
since it's perl, I would expect it to be incomprehensible
cat /dev/random
>>53745844
I want to look at it.
This is like those times you wake up in the middle of the night, overcome by the desire to see what's up in /d/. Like those times you were genuinely curious to know what would happen if you actualy tasted That Guy's food. Like those times you smell your finger after scratching your ass.
I need to see those perl scripts.
>>53741995
This will be the future for python and ruby in 15 years.
>>53745920
I'm assured by the implications for job security.
>>53741805
Post scripts or OP is a fake
>>53741805
who cares, it works better than gnu trash, has a perfect file-picker and runs all gaymes and professional software. OP BTFO
>>53741805
>Some random screenshot of random shit
Nice evidence retard.
>>53746305
Honestly this is the truth.
Even if the screenshot is true (highly doubtful) the fact is that Windows actually works better than anything else out there.
>>53746305
What the fuck is it with you and file picker pictures in like every thread on /g/?
>>53746305
>it works better than [insert anything here]
>>53746305
>perfect file picker
Dolphin would like a word. Seriously, I wish there was a windows port of dolphin.
>>53743655int foo()
{
return bar;
}
Triggered as fuck. What bloat.
>>53746432
It's too bad dolphin is tied to kde so hard or I'd absolutely use it.
>>53743584
#define EXIT_SUCCESS 0
It's something other than 0 on some fringe systems
>>53746432
>>53746802
what's so bloody great about it? it looks pretty... generic?
>>53741805
>They still use the perl scripts.
The entire internet runs on C programs cobbled together with Perl.
>>53746802
kdebase-dialog too
>>53742420
Sounds like the Linux kernel.
>no breaking userspace ever
Fucking Torvalds.
>>53742289
Make a new windows thats really just Linux with a windows skin. The devs will then make every windows application work perfectly on some proprietary wine fork
>>53742936
>so they use flat design as an excuse to make everything boxes so the program managers can make shit in powerpoint and give it to the programmers
I'm finding this hard to believe. Its like Kruger Industrial Smoothing from Seinfeld had a software division
Is he describing Qt?
>>53746897
They basically should make a new system, have compilers make binaries both for NT and the new system, have a new version of Windows that uses the new system only for running the new binaries and does everything else with the old codebase, then 8-12 years later pull out a new Windows that completely kills NT and is just the new system.
Basically transition like NT transitioned from DOS/9x shit. Or sooner or later the weight of decades of legacy shit crushes you, and given the failures of Windows 8 and 10 that time is now.
>>53746996
kek, is there any huge project that isn't a total mess anyway?
>>53746813
>judging by looks
>not by functionality
Wew.
KDE and Dolphin are what you use if you want a file picker with a thumbnail grid. Plus other useful features.
>>53741805
>no one could decipher what the flying fuck was going on with the perl scripts
Hire some Perl neckbeard. Stupid fucking memers.
This is why i switched to Mac OS X 4 years ago
>>53743735
That was DOS you dumbwit
>>53745393
>>aiming for compatibility with Server 2003.
>The Windows version that ReactOS officially targets has been a point of confusion for some time now. There are actually two specified targets, one involving the NT kernel and the other for the Win32 subsystem. Officially the ReactOS kernel still targets the Server 2003 kernel, also known as NT 5.2. This target changes fairly slowly. On the other hand the Win32 subsystem target is always at the latest Windows version released
>>53747104
best thread on /g/, thanks OP
>>53747129
And? A few here and there failures are way better than an OS whose own developers aren't even sure how it functions
>>53747129
> that code
So, since we have multiple MS dudes in here, any word on what goes in in the XP legacy program? What does it look like on the inside; the OS itself, I mean?
I'm curious as to how much it has diverged and/or merged with Vista-10.
I used to work on the xbox division, all I know is that windows works because black magic.
There have been attempts to fix it or replace it, all shutdown by management.
>>53742936
Man, I call bullshit on the Metro bit. Windows 10 UI maybe (but then again, that's no longer Metro), but Zune or early Windows Phone was fucking good.
Like, all of you, go and try out the Zune Music app, that's minimalism done right.
>>53742379
>yfw "NT kernel" is just a sophisticated DOS program running on a DOS 1.0 / 8086 virtual machine
>>53747022
They should pull what OS X was originally going to be.
Make a new OS with good APIs, then Blue Box Win32 for the old shit.
>>53741805
All this is probably bullshit, ms has multiple teams working in parallel producing multiple build per day and then mefing them. If that id stuff is tue, this wouldn't work. I call BS.
>>53747361
>>53747361
this is true, isn't it?
>>53747513
Nope, not after win95, which was the last OS that was running as a DOS application inside DOS.
>>53741805
Why didn't they made the new one from scratch?
They already shat on everything the old ones were and did, so why not starting completely over?
>>53747710
Because OP is shitposting.
>>53746815
Everyone knows the internet is fucked, just like everyone knows that web2.0 was a mistake and webdevs are shitheads.
It's more fun when the blame can be leveled at a single idiot or small group of idiots. Now shut up and enjoy the thread.
all this fucked up mess with sources and compile process and on the top of everything it's still million times better than loonix
>>53747823
but it's not. right poonjab?
>>53742803
I'm calling this bullshit, haha. Too obvious.
Microsoft BTFO: the thread
Look, I despise microsoft and their latest iteration of a broken mess that is windows, this time even a fucking beta test that's still going.
But that's a pile of text written by someone on the internet. How is this any relevant or credible?
>>53742936
>because no one at microsoft knows how to design UI
They used to, tho.
>>53747970
It seems to explain a lot of the issues with Microsoft in its current state, and it seems pretty hard to make this shit up desu senpai
>>53742967
At first I thought this is legit, but the more I read, the more the bullshit is starting to shine through, desu.
>>53747994
I don't know, I just find that people are too reactionary for something that could be completely false.
I mean what anyone can see is that windows 10 is a broken inconsistent mess, kinda like they didn't really touch Windows 7 much but they put shit on top of it.
I wish Microsoft started over and removed all legacy code. I'm sure we could get 50% more performance out of our computers with no legacy code.
>>53748079
but, can we actually do this?
also, we already have a free as freedom kernel and OS, why don't we work on it and make it better than MS Windows?
genuinely interested.
I know tons of people who worked at MS for many years. The reason Windows is such a mess has less to do with their developers but more to do with management. You have people who have no clue how software works giving orders to people who do. Then they ask developers to do the impossible while not breaking compatibility. In the end you get the mess there is today where developers are working with 20+ year legacy code on an uncommon language with no comments.
>>53741805
>nobody knows how it works
it's fucking XML file, everyone knows how it works.
you want to start about deprecated software open sores? you really want to go there?
>>53748079
why would it increase performance of computers that don't use legacy code?
Leaving legacy code if code was written properly in the start just means everyone is using what he needs.
>>53748106
As a wincuck, Linux already is better fundamentally. It just has little to no adoption rate.
>>53742627
>autoconf/automake which actually are insane
>M4
>Everyone knows M4 from all the Sendmail config they write I guess?
>checking for stdio.h.... yes
>detecting if compiler supports -o flag... yes
>checking for printf() .. yes
>checking if OS is Windows 3.1... no#ifdef HAVE_PRINTF
printf("Output goes here\n");
#else
/* Implementation of printf for platforms that don't support it */
asm(...
#endif
>mfw autoconf
>>53743557$ man something
Something(2) manuals are now available in the info pages. Please run info something.
>>53743775
Best program description in history.NAME
true - do nothing, successfully
So what does this mean? Is the core of windows so fucking old no one knows how to handle it and people just kept adding and adapting the new shit to run on top of the old shit? Almost like a VM within a VM etc?
>>53741805
Microsoft confirmed for aedeptus mechanicus
>>53747255
>yfw Gates sold his soul to the devil to make Windows work and become rich
>if he didn't do that it would all be just another spaghetti code base that doesn't work
>>53748170
>if code was written properly in the start
That almost never applies to legacy code.
>>53743193
As someone who recently switched from 2012 to Ubuntu the difference is immense. 1gb of updates finishes miles quicker on Ubuntu than winblows.
>>53748079
If you want a crippled OS that can't run three year old software you can always use OSX
>>53742255
[8:01:19 PM] <Anon> then again one of them sent me example code as a screenshot embedded in a word doc so perhaps my expectations are too high
I still need someone to explain to me what the fuck they did between Windows 2003 and Windows 2008 to make the event viewer go from "snappy" to "I'll go make a coffee while it loads".
>>53748175
>we started using centos
>a fucking enterprise OS
>years passed
>in recent months they removed firewalld from centos 7.2, just like that, it wasnt even in the changelog I think
>had to manually fix scripts at hundreds of servers.
thanks free software, i guess
>>53748667
I still think if they made it a requirement to use curly braces around all statements then this couldn't have worked in the first place.
>>53743296
It's the fucking shazbowl all over again
>>53743087
I feel like this is every major company now. Management nowadays are the ones preventing any kind of positive change and think they know how to handle their dev and tech teams all while having zero experience in those fields.
ITIL was the worst thing to ever happen to the tech world and computers nerds.
Moral of the story: never use Perl or you will be trapped forever.
>>53746305
my gnome desktop has a file picker with preview image i dont know what you talking about.
(ubuntu gnome 15.10)
>>53741805
>>53749275
Fucking hell
>>53749275
Thanks a lot for the compilation, Anon, you're golden!
>>53749275
Memecrosoft
>>53742381
>I think they cut out a sizeable chunk of it in 8/8.1 because no one used it.
I doubt it.
You can still install MS Word 95 on WIndows 10 OOTB with no issues.
>>53748746
Wait, wait, wait.
Wasn't firewalldicks advertised as THE replacement for iptables scripts, because people were to dumb to use iptables-restore/save instead of custom, insecure bash shit?
>>53749275
saved.
I remember reading an extensive article from a Windows Kernel Developer saying he was jelly of Linux Kernel because Windows Kernel was bad and every developers felt bad.
If only there was a REAL alternative to X for desktop GNU\Linux GNU\linux could be the greastest desktop OS ever made.
So Windows is built on shit code layered on top of more shit code just so that the first shit code doesn't break any of the old shit code.
Damn. Microsoft sucks.
>>53749111
Trips of truth.
Management and MBAs are the what eventually drives a company to shit.
>>53749466
>If only there was a REAL alternative to X for desktop
There's about to be several.
GNOME3, KDE5, Cinnamon(eventually), MATE(eventually), XFCE(possible at least). If these DEs ever finish their wayland support then they wont even need X for anything except X applications.
