Re: [dev] Interesting post about X11

On 15 Jun 2010, at 23:33, Connor Lane Smith wrote:

> On 15 June 2010 14:05, Ethan Grammatikidis <eekee57_AT_fastmail.fm>
> wrote:
>> On 15 Jun 2010, at 12:51, Connor Lane Smith wrote:
>>> In my opinion the problem is purely user experience:
>>
>> Is this your opinion, or lines you've been fed?
>
> My own, strangely enough. UX is one of my interests, particularly
> since most programmers seem to suck so hard at it. Let's at least
> assume for the time being that suckless devs have minds of their own.

How strange, mine too. I've been "interested in" (read: "painfully
aware of") user experience issues ever since my first Atari ST. I've
battled with every post-8-bit computer I've had on user interface
issues, where your opinion appears to coincide with some fairly
popular ideas I've found the hard way to be harmful. I'm sorry if I'm
misinterpreting your opinion.

>
>>> (a) installing
>>> software is perceived as difficult, so not having to bother with
>>> that
>>> is an instant plus, and (b) your data is available everywhere.
>>
>> Neither is a reason to use horse-shit like HTTP, HTML, CSS or
>> JavaScript.
>
> You clearly didn't read anything else I wrote. Just for you I will
> repeat myself:

Actually, I did. I went back and read it 3 more times before replying
here. The first time around I thought you were sticking up for web
apps and the "web-based OS". Apologies if I was wrong, but I'm still
not sure that's not what you meant.

>
> On 15 Jun 2010 12:51, Connor Lane Smith <cls_AT_lubutu.com> wrote:
>> However, HTML and HTTP do not form a panacea. It's just the "in
>> thing"
>> nowadays to write slow, buggy clones of existing software in
>> JavaScript and to call it innovation.
>
>> Both of
>> these problems could be solved with improved package management and
>> data synchronisation, but it's an uphill struggle, since everyone's
>> just fixated with the intertubes now...
>
> How tiring. Next time I suggest reading what people actually write
> before replying with snotty shit.

"Improved". You can't take a broken concept and improve it. That way
lies the path of suck. As far as I can see, the notion that you can
take broken concepts and glue "improvement" onto them is THE path by
which the software world has become so completely flooded with 'suck'.

Package management? You're ignoring history, a lot of it. Linux distro
people have been struggling to make package management work for 18 or
19 years now and they still haven't made it "easy" enough. Lindows/
Linspire had one-click installation years ago. I never used that, but
I tried Ubuntu 6 months ago which had a beautiful package management
user experience. I'm not being sarcastic, it was really nice. It
wasn't enough. I had a hard job to find the right versions of some
packages for a particular commercial app I'm stuck with, and that app
already had a whole bunch of pre-built libraries to ease compatibility.

Data synchronisation? If anything this problem is worse. The issues of
synchronising raw data are solved. rsync, numerous revision control
systems, venti, and I'm sure many other technologies all offer more or
less efficient ways to synchronise two or more stores of data. There
are two hard problems: making the data meaningful to both ends and how
and when to synchronise.

Making the data meaningful to both ends is not a problem if both ends
have the same versions of the same software. Otherwise you're talking
about a standardised data format, and standards for end-user data
don't seem to work very well. There's always something which could
reasonably be added.

How and when to synchronise is a harder problem. I could set up my PDA
so I push a button and it synchronises with my desktop wherever in the
world I might be, but what happens if I've changed data on both PDA
and desktop separately? It could be fixed with user interaction, to be
sure, but who would want to deal with that? I know of one system which
attempts to handle such a case somewhat automatically, Plan 9's
replica, but it doesn't work well enough. Plan 9 users have lost
critical components of their operating system to replica, more than
once. How long has the need for synchronisation been around? How old
is rsync? 15 years, not quite as old as Linux package management.

Do we have a solution yet? No. We might get one tomorrow, BUT, and
here is something I wish all the "improvement" fans would recognise,
*expecting* research to come up with what we want is stupid. Research
is finding out things, it's not necessarily going to give you what you
want at all. I don't have any faith that any solution to any problem I
can see may be discovered any time soon. I *am* willing to do research
myself, but I an VERY careful about introducing hypothetical solutions
which may never work out. It's not just noise, it actively leads
people to make harmful software. I think that's how SVN got the way it
did.

Another bad habit which is too common these days is taking a pretty
idea and never examining it to see whether it is in fact worth
anything. When thinking over the synchronisation problem today, for
one whole minute I actually believed web applications were a good idea
after all. And then I remembered my own experience using my own
computer from the other side of the atlantic via a hotel's computer.
At the time I had a 64kbit uplink. the protocol was vnc. It worked
fine. There was a slight delay when moving windows, over a 64kbit
link! Did I mention my home machine was running Enlightenment E16? The
window manager with possibly the shiniest and certainly the most
detailed themes ever seen (the one I was using was no exception), over
a very detailed background, on a 400MHz AMD K6 CPU (the K6 was crap),
using a standalone X+VNC server (no hardware acceleration of any
kind), over a 64kbit link to a *Java* viewer, and the biggest problem
was a _little_ bit of hesitation!

There's no reason to subject application developers to the horked-up
mess that is web "technology". It's no solution to anything, mediocre
interfaces are all over the web. As a user I don't want to be
subjected to the crap I see on the web in my applications! If you want
remote access, from any internet cafe and half the libraries in the
world... How convenient, never worry about your laptop getting stolen!
If you want such remote access get yourself a remote framebuffer
viewer which works in web servers and get the web crap done once and
out of the way! There's one such viewer already, in tightvnc, with a
tiny httpd to serve it and all GPL iirc, or I can imagine something
quite simple could be done with the html 5 canvas and a little
javascript pulling from the server. If you want high availability get
yourself a VPS and run your apps on that.

I find it slightly amusing, slightly sad that my old VPS provider has
had describe themselves as providing cloud services, or something like
that.

-- 
After watching Linux go nowhere for 10 years, actually watching it go  
backwards many times, I woke up.