Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon.

Pages: 1-4041-

Wirth's law

Name: Anonymous 2016-01-11 13:34

"Software is getting slower more rapidly than hardware becomes faster."
-- Niklaus Wirth (February 1995). "A Plea for Lean Software".

I think this is something we here on /prog/ can clearly see. Programmers keep making things more and more complex/bloated shit for no good reason. Almost nobody cares about lean, fast, responsive, optimized software. Sad times.

Name: Anonymous 2016-01-11 13:39

>>1
I agree but there are others who see this.

Name: Anonymous 2016-01-11 13:42

Name: Anonymous 2016-01-11 13:46

How do I join?

I share a lot of their views and stuff but like, what do you do?

I don't think they really need me is the thing, they're getting on pretty good by themselves.

Name: Anonymous 2016-01-11 13:47

also suckless are a little too close to cat-v (e.g. one of their guys runs the site, they have some connections to that IRC channel) which is very offputting. But I don't mean to judge them..
cat-v is really disgusting though.

Name: Anonymous 2016-01-11 13:54

>>4
Find something you like on http://git.suckless.org/ and check out the TODO file.
Or write good software following the philosophy and ask for it to be added there so others can benefit and help you out with development. That's what many do.
Also check the latest conference for some nice talks: http://suckless.org/conference/
The best IRC channels are #2f30 on freenode and #suckless on oftc.

Name: Anonymous 2016-01-11 13:54

>>1
GC, Lazy evaluation, dymamic typing and late binding
are not considered luxuries but normal.

for no good reason
Software complexity is mostly caused by user demand:
example CSS rendering(99% of which is useless
animations/effects) which is required for web
browsing. A simple no-css no-js browser would
be much simpler in design, but satisfy much less
users.
Another point is language used for performance
(C++) is not geared for
optimization like C, its filled with hidden costs
that manifest if any optimization fails.
Another common source of bloat is templates,
templates are generated for every use-case compiler
cannot safely ignore and are resolved very slowly
(many nested templates specialization is practically
exponential). Using many .c and .cpp files to produce
object code(often redundant) which is then optimized by link time optimization
is far less efficient than one single program.c file, and
lack many optimization opportunities, besides the
template mess such approach produces for the linker.

Name: Anonymous 2016-01-11 13:59

Suckless isn't the only ones out there.
I particularly like Quentin Carbonneaux's software.
http://c9x.me/
But he's also a regular on suckless mailing list, nonetheless.

>>5
They just share the idea of keeping software simple, but I don't think the relationship between them is too tight. I also rarely see an intersection between suckless and cat-v people on each other's irc channels, mailing lists, git repos, etc.

Name: Anonymous 2016-01-11 14:09

The R-36 (8K67) ballistic missile, known in the west as the SS-9 SCARP, was a two-stage, tandem, storable liquid-propellant intercontinental ballistic missile.

http://git.r-36.net/

Name: Anonymous 2016-01-11 14:16

>>7
Useful features are needed, but they should be implemented in the simplest way possible. What we're talking about cluttered implementations VS simple implementations (not non-implementation).

See web servers, for example. Let's say... Apache and nginx. Both have more or less the same features and work to serve web pages. But they solve it in different ways.

Want to go to a lower-level? Check out libc implementations.
http://www.etalabs.net/compare_libcs.html
glibc is what almost everyone uses, but it's horribly bloated, slow and error-prone. uClibc and musl solve the same issue glibc does and have the same features, but they are much better designed, with simplicity in mind.

We aren't talking about ditching (useful) features, but about implementing them in a good way.

Name: Anonymous 2016-01-11 14:23

If anyone could explain rationally
the point of suckless besides "no bloat"
Anyone could write a few dozen of the simple
C programs PER DAY.
They're primitive and just expose libC/OS functions
to end-user with very little original code.
What exactly they need? A minimal X is also
the less useful X. If e.g. you limit codebase
to 10K lines, you'll start missing useful features
because they're "inelegant" or "use too much code"
like if we lived with 1MB of ram.

Name: Anonymous 2016-01-11 14:31

>>6
List top 3 progs they(suckless) need and their specs/functions.

Name: Anonymous 2016-01-11 14:34

>>11
Anyone could write a few dozen of the simple C programs PER DAY.

Maybe you think that because they are small, but a lot of thought goes into thinking about the best way to do it. Check out st, for example. http://git.suckless.org/st/tree/
That's clearly not a one-day project, you are undervaluing their work.

you'll start missing useful features because they're "inelegant" or "use too much code"

There's a trade-off between "useful" and "complex". Is the feature useful enough that it justifies making the codebase more complex? If it does, the feature is added. If it doesn't, it isn't. We can't have it all.

Name: Anonymous 2016-01-11 14:42

I made my own version of top. it's 1/10,000th of the size if you count lines of code.

what do those extra lines do/mean? I'll tell you one thing they bring: bugs.

Name: Anonymous 2016-01-11 14:45

>>13
In that case,the "best way" could be figured in
minutes, not days. What exactly suckless
debates on? Elegance or code size?
simple terminal is dependent on X11 - does X11 seems
elegant or small to them?
>>14
C code without discipline invites bugs.
Modularize and unit test everything.

Name: Anonymous 2016-01-11 14:48

>>12
If you're looking for something simple to start off, sbase has a few missing tools. http://git.suckless.org/sbase/tree/TODO Should follow POSIX 2013.

smdev is being considered as Alpine Linux's default device manager, instead of the horrible udev. It could use some help too. http://git.suckless.org/smdev/tree/TODO

There's also the simple dhcp client (sdhcp) that's far from completion http://git.2f30.org/sdhcp/file/TODO.html

Basically, just look at a program you like and find useful for yourself and check out the TODO.

There's also stali, their static linux. http://sta.li/

For brand new projects, I don't know what exactly they need.
The best would be to ask that on the mailing list (remember to disable HTML on your email client, only plain-text email is received).

Good luck!

Name: Anonymous 2016-01-11 14:54

>>15-san brings a good point.
Is dwm suckless? Yes, it's a suckless window manager looking at the code. But it depends on X11 wich is hugely bloated.
Is surf suckless? Yes, it's a suckless web-browser front-end. But it depends on webkit which is hugely bloated.

But this is understandable. There isn't enough man-power to work on huge stuff like an entire browser engine or windowing system by themselves. This is an inherent issue.

All in all I'm glad to use X11+dwm instead of x11+GNOME for example. It's a start.

>>15
In that case,the "best way" could be figured in minutes, not days.
Members have jobs and other responsibilities. Maybe once the community grows development can be as fast as you want, but we can't blame them right now for the slow pace.

Name: Anonymous 2016-01-11 15:02

>>15,17
They discussed a possibly migration to wayland if, after through research, it proves less complex and more performant than X11. But it's still soon.

Name: Anonymous 2016-01-11 15:07

lets make a suckless OS on top of a simple computer architecture and instruction set

Name: Anonymous 2016-01-11 15:10

>>19
lowRISC by the Cambridge University people. http://www.lowrisc.org/
But your proposal is completely outside our scope. The amount of time and energy needed for that could only be accomplished by sponsors. We have jobs and working on an operating system requires full-time focus.
Out of question.

Name: Anonymous 2016-01-11 15:15

>>14
Can you share?

Name: Anonymous 2016-01-11 16:06

Who is cat-v and why are they hated?

Name: Anonymous 2016-01-11 16:16

>>22
It all started here: http://harmful.cat-v.org/cat-v/
Today cat-v developers work mostly on 9front.
They are hated by some people probably because of their "in your face" style. I'm personally okay with cat-v, but it's something very out of the ordinary.

Name: Anonymous 2016-01-11 16:27

>>23
Okay, so they're still the same shuffling of deck chairs on the Titanic as "suckless" is doing.

The problem is Unix, processes, hierarchical filesystem, and streams of bytes. The notion of an OS needs to be rethought if we're going to scale past 1970s computational architectures and failings, because attempts to make up for those failings are bringing in all the buggy crap as we try to make computers more useful to more problems.

Name: Anonymous 2016-01-11 16:31

>>24
I fully agree with you, but such a thing is far-fetched and would need hundreds of people working full-time to accomplish.
In the meantime, I'm happy to have dwm instead of gnome, st instead of xterm, musl instead of glibc, sbase instead of coreutils, etc. It at least minimizes the harm to a certain degree.

Name: Anonymous 2016-01-11 16:39

>>24
I have to agree with >>25-san.

Rob Pike shares your idea and so do many of us. But your proposal is something that doesn't happen over night. For now, we better work with what we have.

Name: Anonymous 2016-01-11 16:55

>>6
Does suckless need any help with tech evangelism or drawing up a code of conduct?

If so I can help.

Name: Anonymous 2016-01-11 17:10

>>27
No, just programming help.

Name: Anonymous 2016-01-11 17:21

>>25
"rotten foundations"

Name: Anonymous 2016-01-11 17:31

>>29
Better "rotten foundations" than "rotten everything".
Of course a suckless windowing system would be great.
Of course a suckless browser engine would be great.
But with the current manpower limitations, that's unattainable.

Name: Anonymous 2016-01-11 18:21

Don't forget aiju!
I like his Chem program.
http://aiju.de/code/chem/

Name: >>31 2016-01-11 18:25

Name: Anonymous 2016-01-11 18:28

Is this the reddit expat thread? We don't have to be told about this.

Name: Anonymous 2016-01-11 18:33

>>31
I don't. Beside the batshit DSL, why require the angles for each edge? Can't it figure that out from the number of connections on each node?

Name: Anonymous 2016-01-11 18:33

>>33
We're just talking about simple/lean software. Chill out, fella!
By the way, you think you're great and all, but you don't even know how sage works. Lurk moar.

Name: Anonymous 2016-01-11 18:34

>>35
>>33 presents a typical kneejerk reaction to >>1. Please ignore.

Name: Anonymous 2016-01-11 18:37

>>34
I'm no chemist, but I think it could take different angles?
Also, you could want to represent it like this http://www.chem.wisc.edu/deptfiles/genchem/sstutorial/Text7/Tx73/AcOH.gif or this http://www.chem.wisc.edu/deptfiles/genchem/sstutorial/Text7/Tx73/AcOHa.gif
There could be a flag for using drawing one or the other.

Name: Anonymous 2016-01-11 18:39

ACME's web server has always been a personal favorite in the past. Who remember thttpd?

Name: Anonymous 2016-01-11 18:44

>>34
That's the rub, isn't it?

Should software be simple & lean, or should it have niceties added? A nicety for one might be essential, while to another is just bloat.

Name: Anonymous 2016-01-11 18:44

if suckless was serious about minimalism
they would switch to freedos
and make dos programs.

Name: Anonymous 2016-01-11 18:50

>>39
Exactly!
There's a trade-off. Complexity VS Feature.
Is the feature of "deal automatically with angles" useful enough to justify adding a bit more complexity to the code base? Well, I guess it is! Because otherwise it's a big hurdle to do it manually. But adding a feature of, let's say, "output image in png, jpg, bmp and pdf" a good one? Probably not. Better keep it simple with a single output and you convert yourself if necessary.

The trade-off is not always so easy to reason about. But I think the suckless guys do a good enough job (from the tools I've used so far).

And minimalism is not only about features, but equally (if not more) importantly about how a feature is implemented.

Name: Anonymous 2016-01-11 18:52

>>40
Suckless isn't about only minimalism.
It's about the trade-off between usability and minimalism as seen on >>41
Wanna go fully minimal? Then stop using the computer.
What suckless seeks is a minimal solution to computing, while keeping the most important features.

Name: Anonymous 2016-01-11 18:58

>>39
Is a better input format necessary? Do I have to write a program myself just for the purpose of translating a more usable format to the one this program accepts?

Those are the question of feature vs. bloat.

Name: Anonymous 2016-01-11 19:52

>>42
But for what purpose? To make your multigigahertz CPU work less? To make this more manual for you?

It might be a tradeoff between bugs and features, but I doubt it. This is a small team building new codebases with little time & use for maturation.

Their time would be better off doing things a new way, rather than rolling back the clock on today's software, which is effectively what they're doing.

Name: Anonymous 2016-01-11 20:07

>>44
``a new way'' is the way people have been trying to do things for over half a century.

Name: Anonymous 2016-01-11 20:29

>>35
you don't even know how sage works. Lurk moar.
Epic win /b/ro!

Name: Anonymous 2016-01-11 21:38

>>1 think about it as demand-creation:
Without new software requirements the hardware would
not exists. Imagine a world with lean,hyper-efficient
software with minimal features, but same old hardware.
The cost to create new chips/videocards/cpus cannot
be justified until there is demand for it.
That where bloatware comes in, it can barely run on
current hardware, but some newest expensive system
can do it without a hitch. Suddenly this expensive
system is in demand, and companies begin targeting
bloatware market with cheaper system by competing on
price and features. Supply follows demands.
Intel wouldn't even exist if not for an obscure
calculator company they got to design a chip for.

Name: Anonymous 2016-01-12 0:34

>>45
Most of the half century has been trying to extend old ways. People stopped making new ways somewhere in the 1960s.

I guess the drugs helped? (but seriously, the lack of prior art meant they were unconstrained, unlike today)

Name: Anonymous 2016-01-12 0:42

>>47
Are you 18 yet, FrozenAnus?

Name: Anonymous 2016-01-12 5:29

>>36
I visited reddit yesterday. It was (still) terrible
did a quick nsfw search, because you know, just because..

mouths being used came up, and i thought, uhh, ok..

Have you seen how redittors use their mouths? It's no wonder i cannot post there

Name: Anonymous 2016-01-12 5:41

>>49
Age is just a number.

Name: Anonymous 2016-01-12 5:51

=) major lazer - free
Reminds me of Mr Jitendra Malik ^^

ugh, i was supposed to be studying Crypto II with Dan Boneh about now =.

Name: Anonymous 2016-01-12 6:22

Probably my bad for distracting him with that factorial based crypto in crypto I ^^ c'mon danny bones!

P.S Redditors should make their way to lounge and rinse thoroughly with redcream

Name: !fig9A 2016-01-12 7:06

I don't think i'm going to get along with these redditor expats..
I'm too #nofilter for reddit and 8chan alike ^^

being non-pro-american and non-pro-asio was too much for either of 'em xD

Name: !fig9A 2016-01-12 7:17

And i like obama, but it's a bit like putting a smartie on top of a dog turd, for the most part it's just a waste of a smartie

Name: !fig9A 2016-01-12 8:07

Silly americans, what if instead of needing guns to protect you from a dud government, you just got rid of the dud government?

Name: Anonymous 2016-01-12 18:00

>>56
That's racist.

Don't change these.
Name: Email:
Entire Thread Thread List