Re: OT: my new PC rocks!!
From: Beth (BethStone21_at_hotmail.NOSPICEDHAM.com)
Date: 12/30/03
- Next message: Beth: "Re: OT: my new PC rocks!!"
- Previous message: Madhur: "why MASM generated this code"
- In reply to: Frank Kotler: "Re: OT: my new PC rocks!!"
- Next in thread: pH: "Re: OT: my new PC rocks!!"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 30 Dec 2003 15:55:20 -0000
Frank Kotler wrote:
> Beth wrote:
> > Randy wrote:
>
> >>Windows is the only OS I can currently install on the machine as
the
> >>hardware is too new for most of the other OSes I've got (no, I'm
not
> >>interested in installing drivers by patches, recompiling the
kernel,
> > etc.).
>
> > Mind you, I _still_ can't get Linux installed on my "Linux box"
second
> > machine...
>
> I can get Linux installed, no problem. (my "first machine" is made
of
> spare parts, but I guess they're sufficiently "standard"...)
Aaarrggh! Will people stop saying that!! I mean, when someone comes up
to you and says "I'm not feeling too good, I think I've got a bug and
am low and lacking in energy and feel miserable", is the best way to
comfort them to jump up and down in Teletubby "happy, happy, joy, joy"
Utopia, nah-nah-nah-nah-ing how everything is brilliant in your
universe and you're not ill at all? Know what I mean?
Yes, I _know_ Linux can install no problems...it _used to_ do that for
me too...but when it stops doing that, then you'll also go around
cursing those "evil" ISO images and all those people smiling like the
Cheshire cat in their happy, singing "Toon town" universe of pure
joy...
> Getting Xwindows configured is another story! I don't have the
original manual
> for my monitor (secondhand - my last monitor literally "fell off a
> truck"!) - maybe if I googled around I could find specs online,
but...
I'm alright with this because I kept the big monitor box (because it's
nice and big :) for storing things in...requires climbing into the
attic with a torch but the specifications are nicely written in big
letters on the side of the box to type into "Xconfigurator" directly
for perfect monitor perfection :)
> Windows doesn't ask me all those tough questions, why should
Xwindows
> have to? I assume the answer is "better hardware probing". Maybe,
just
> maybe, assembly language programmers ought to be doing something
about
> this instead of complaining about it.
Well, there may be a bit of "better hardware probing" in the sense
that Windows may recognise "plug and play" monitors better...but,
really, I don't think the problem is that at all...
You know what the real difference here is? _Attitude_...think about
it...when you boot up Windows, it's usually set to the lowest
resolution with a reasonable amount of colours and the refresh rate at
one of its lower settings...then, you go into "display properties",
move the sliders up a notch, press "test" and see if the screen goes
all wobbly (and, yup, it sure can do that :)...if you don't press
"yes" in 15 seconds, it reverts back to "safe"...well, try moving the
notches to less extreme settings...then try again...once you've got it
right, you never go back...and, hey, you might even forget that you
spent that time playing with the settings to get them right...
Windows determines the settings "empirically" with the "interactive
assistance" of the user...it starts "simple" and then the adventurous
user moves the notches higher...X-Windows, though, wants to know the
_exact numeric values_ before proceeding...its "attitude" is to use
those numbers to work out what's possible with formulas and stuff...
The difference is merely "attitude"; If X-Windows simply chose
"640x480x256" and then featured a set of GUI sliders where you could
"notch" it up to higher settings, complete with the "if you don't say
'yes' in 15 seconds, then we'll presume the image went all 'wobbly'
and it's not a valid setting" thing then you'd find no difference
between Windows and X-Windows on that score...
It's two different solutions to the same problem; And though
X-Windows' solution is _logically_ "superior" because it shouldn't
even offer the choices of settings that'll never work (presuming you
give it the correct numbers :), the Windows version is _physically_
"superior" because it has "hidden" those questions behind moving
notches and a simple "yes or no" as to whether you can actually see
anything on the screen or not...well, any "fool" can move a notch and
then say "yes" when it works and "no" (or just leave it 15 seconds, if
the screen is so mangled that there's no way you can tell where the
mouse pointer is to move it to the right button ;)...
Windows takes a different attitude: "Hey, those modern monitors can
handle being given wrong settings for 15 seconds or so...let the user
find the best setting that actually works by simple trial and
error"...my video card can do an awful lot more resolutions and
refresh rates than my monitor can handle but I know I've got the best
resolution that it can handle because I moved the notches up one by
one until they were as high as they could go without the image going
all "wobbly"...
X-Windows really needs one simple-ish addition (which was originally
absent from Windows, until Microsoft wised up :)...the ability to
"reset" the resolution / colour depth / refresh rate settings
on-the-fly...that is, change the video settings without needing to
completely shut down X-Windows and reboot it...as I say, in Windows
3.x days, you also had to select your video settings and driver then
shutdown Windows completely and reboot it to see if those settings
would work...that was as much a nightmare to use as X-Windows still is
to configure...
But if there was some "SetVideoSettings(width, height, colour depth,
refresh rate)" API that immediately changed video modes - resetting
the desktop to this new resolution - then it would actually be pretty
easy to write the "notches" GUI interface that allows the user to test
out each setting to see if it works...then, if they click "apply",
write the settings to the usual configuration file...
Plus, for "full-screen" computer games and so forth, X-Windows could
do with stealing a little of that Windows functionality with some
"SetVideoMode" API thingy...as I say, Windows was itself once exactly
like X-Windows on this that it was sort of "stuck" in one resolution
that wasn't easily changed...but Microsoft added the required video
mode changing API and then adding the "Display Properties" utility is
child's play after that...
Windows doesn't actually do a better job when you really consider
things...what it does, though, is "hide" the fact that you're even
doing these things...I mean, you _must_ have played with "Display
Properties" sometime but you do it once and it's so easily done,
you've not even realised that, effectively, you _DID_ answer similar
questions with Windows too...the trick was changing the questions from
"please enter the vertical sync rates" into sliding some notches
instead...and, sure, we _allow_ for invalid settings to be chosen but
we only keep them when the user says "yes, please keep them"...
Microsoft, to their credit, have - though only through "trial and
error" from a load of user complaints, I'm sure - learnt that
sometimes the "techno-fix" is the _worst_ possible way to fix a
problem...sometimes, even if not quite so "impressive", a simpler
solution ends up better because it's easier on the user and the
"drawbacks" really are completely tolerable (yes, Windows does allow
the image to "go wobbly" but that's not really such a terrible thing
to get quite so worried about avoiding it...modern monitors - thanks
to having to handle that Windows takes this approach, in fact - will
be able to handle bad settings for a short period of time without any
great trouble...some, in fact, will simply automatically switch
themselves off when given bad settings so you can't cause any damage
:)...
> > ...I checked the jumper settings and they
> > are okay...
>
> I assume you're talking about "master/slave" settings? (pardon the
> politically incorrect language) That's a good start. I've had "bad
> jumpers", too, that appeared to be in the right place but weren't
making
> a connection. "Wiggle the cables" is a good high-tech fix for many
ills :)
Well, the "jumpers" are correct...that much I've made sure of...and
I've connected and re-connected the wires a few times (taking things
out, putting things back in, etc. :) that, though I didn't
specifically try "wiggle the cables", it would be unlikely that I'd
repeatedly make exactly the same mistake each time...anyway, when I
plug in a cable, I usually naturally "wiggle" them a bit to make sure
they went in correctly...
But it's a very good suggestion...Murphy's law _does_ often tend to
rule in these cases and something that simple is often the
problem...but not in this case, I'm sure the jumpers and "cable
wiggling" is exactly right...I don't take offence at the "easy"
suggestions because I've had problems plenty of times that ended up
as: forgot to plug it in, loose wires or something equally obvious and
silly...
In fact, I had a problem with a floppy drive once...it kept having
"bad sectors" at a certain area of the disk...I couldn't work out what
was wrong...so, eventually, I phoned the "helpline" and the bloke on
the other end simply asked "are you putting the monitor on top of the
box?" (it was a pre-tower PC and, as was customary, I put the monitor
on top of the "base" :)...I said "yeah, sure"...then he said "take it
off...that model's case isn't designed for the weight of a monitor" /
"Oh"...I took the monitor off and everything was perfectly fine once
again...thinking it through in my mind, those "bad sectors" would be
more or less located exactly where the corner of the monitor would
have been pressing down on the floppy drive...
Yup, it was that simple...though it was somewhat "traditional" with
that style of PC case to simply put the monitor on top of it to save
desk space, this particular hardware manufacturer apparently disagreed
with this concept...and, nope, it wasn't designed to handle it at
all...I put the monitor in the wrong place...and that was all that was
wrong...but, as the problems were all with the floppy drive, I wasn't
even looking in that direction at all...the bloke on the helpline knew
it straight away, though, the second I quoted the model number
because, apparently, _everyone_ had assumed putting the monitor on top
was a good idea and this was "complaint #1" when it came to that
model...you can't really call it a "design fault" totally because they
did know but _delibrately chose_ to make it this way...it's more an
"incompatibility" with what users wanted and expected with what some
designer decided "should be the case"...
> > but it still crashes in a recursive kernel panic when I
> > select the "test the CD" integrity test thingy...
>
> What do we know about this CD? Does a Windows "rescue disk" find it
> okay? I'm using Slackware - they provide several pre-built kernels
with
> different support included. If RedHat does it the same way, maybe a
> different kernel would help?
Well, no, Red Hat seem to do it completely differently...at least,
there's only one package available for free download...which, of
course, is how you can say "Red Hat 8" and "Red Hat 9" and so forth
because they concentrate on creating a specific "package"...plus, the
"test the CD" thing I mentioned is actually part of the installer
program and this auto-boots up off the CD with no problems...before
starting the install, the installer has the option of quickly
verifying the integrity of the three CDs to check that the ISOs were
written properly and so forth...much better test it quickly there than
to discover the problem half-way through the actual install, right?
Also, they provide MD5 sums for each of the ISOs, which you can use to
check they downloaded okay and I did check those and the actual images
I downloaded are correct...it might have badly written to the actual
CD but then I've never had problems writing to any other CD from the
same box and I even selected a slower speed for writing just to make
sure it would write okay...it _could_ be that, perhaps, but it doesn't
seem too likely as I took all the expected steps to ensure that the CD
images were correct...
Plus, what happens when it goes wrong, as I say, is that there's a
never-ending stream of text flying down the screen, listing CPU
register contents making it look like an exception or something...and
this is totally unreadable at the speed it's going and ruins the user
interface thing as it scrolls off the top that I'm sure this isn't
_supposed_ to happen...the program itself has crashed (because it gets
stuck spitting out register contents and exception numbers forever to
the screen) rather than this being an "integrity compromised" error
message or something...
And if I choose to "skip" the integrity test then when it comes time
to start writing things to the hard drive, the program freezes for a
second before it automatically reboots itself...which is interesting
because nothing goes wrong reading from the CD for the rest of the
installer (which uses an X-Windows interface to have a nice,
user-friendly install program :)...it's only when it seems to be
attempting to begin to access the hard drives that things go
wrong...I'm almost certain the actual install CD is just fine because
the MD5 sum is right, I took steps to make sure it wrote to the CD
correctly and, with Red Hat, it's an auto-booting CD that starts up
Linux, starts up X-Windows and runs a fancy installation program all
off of the CD and that works 100% fine _until_ it tries to go to the
hard drive...
> > and if I try skipping
> > that, then the machine hangs and reboots itself when trying to
format
> > the partitions ready for install...
>
> Some particular partition it has trouble with? Would resizing them
so
> they fall on different borders help?
Now, there's an idea...I have split up the hard drive into partitions
because I wanted to leave some room for also installing other
stuff...perhaps I should reformat and shuffle those partitions
about...but, again, I wouldn't think it was a problem because I had
something like that before and there wasn't any trouble (plus, Linux
_is_ the first partition because I read about some often reported
problem with larger hard drives because the BIOS only has CHS or
something...so I followed the "read me" advice there that this
_shouldn't_ be the problem and the BIOS _does_ seem to understand such
things because the BIOS set-up program correctly reports the drive and
lists them as "LBA" / "CHS" and so forth...it's not that old a machine
for it not to be able to understand such things in the BIOS ;)...
> > that's Red Hat, by the
> > way...
>
> A long time ago, I heard that if you've got a friend running some
distro
> of Linux, that's the best distro for you. I didn't really have a
friend
> running Linux, but I was talking with the son of an old high school
> friend, and he was using Slackware, so that's what I picked. Never
had
> occasion to use another, so I don't know how much difference there
is. I
> watched a friend install... I think it was Mandrake - seemed to have
a
> fairly "Windows-like" installer... seemed to go well, but the
network
> card didn't work. I think I could have fixed it, but she just
overwrote
> it with something else. Looked like a pretty easy install process...
if
> it had only caught the network card...
Red Hat has a "Windows-like" installer too...plus, as I said, I had
this distro (the previous version of the package, though ;) installed
on this machine before...I was only "upgrading" it to a newer version
(but doing so by re-installing it from scratch, as I wanted to also
shuffle the partitions at the same time :)...I mean, if this continues
not to work, then I might have to simply consider using another
distro...but I was happy enough to stick with Red Hat because they
work in "packages" and have "Windows-like" installers and so forth
that it's _usually_ all very nice and simple...you know, stick in the
CD, answer some questions about your keyboard layout, time zone and
what packages you want then it basically does everything else for
you...a fantastic little install program _WHEN_ it actually bloody
works...my problem here, of course, is that it's not working...and
then things aren't quite so happy or rosey with Red Hat...
> One of my first installs, the network card didn't work. I determined
> that it was using the right "tulip" driver. Went to some nasa.gov
site
> that had the newest versions of the drivers (using Windows, which
may be
> considered "cheating"). Fortunately, the file included the
command-line
> to build it - I never would have figured it out by trial and error.
To
> my astonishment, it worked! "Look ma, I'm a Unix Guru!" But it was
just
> beginner's luck - haven't had any great success since...
Actually, here's a tale that'll please the Linux people...because
having two machines next to each other, as I've set up here, I thought
it would be a great idea to pop some network cards in both and hook
them together...making it easy to move files back and forth, allowing
both machines to "share" the internet and that sort of thing...like
your experience, I got the latest Linux driver off some website, typed
in the command-lines and it was up and running
problem-free...great...but, in fact, the problems started when I
plugged the other network card (same exact card type) into Windows and
got stuck in one of those never-ending "plug and play" loops: "you
have new hardware" / it installs the drivers off the drivers disk /
the driver doesn't seem to be working because it again reports "you
have new hardware" / it wants to re-install the same driver / and
again and again..."okay", I think, "time for Windows update"...yup,
sure enough, there's a newer driver that reportedly fixes bugs (though
the bugs listed don't seem to be the one I was having)...I try it with
that instead...nope, still get the "you have new hardware"
never-ending loop...
I tried everything but Windows simply wouldn't have it...so, regards
the "Windows always installs things easily", this isn't completely
true because, in fact, for this network card, Linux is the one that's
entirely problem-free by just following the instructions for compiling
it supplied in a "read me" file...while Windows - with all its "plug
and play" stuff - simply couldn't cope and, worse, actually gets
forever stuck in a "found new hardware" infinite loop that could
really cause serious problems because the only way to actually get out
of it is to shutdown and remove the card...which is hardly
"user-friendly" to clueless newbie users...in fact, for any writing a
hobby OS with some Windows-like "plug and play" nonsense, there's a
good suggestion for you: have the OS _record_ what it's trying to do
and if it senses that it's tried to install the same piece of hardware
five times in a row, then simply mark it as "not working" and _stop_
trying to constantly "found new hardware" with it...clearly, it's not
working and you've got to have an OS that knows when to give up...it
can be told to retry the hardware manually or something when the user
_really_ does want to "give it one more shot"...but when it simply
gets stuck in a never-ending loop, that's a poor show...
> > downloaded off their site...
>
> I suppose you mean an iso image? With Slackware, at least, you can
also
> get "disk sets" - intended to be installed from floppy, but you can
put
> 'em on a FAT partition and install from there. Might be an option if
you
> can't get the CD working. I guess you can install from an iso image
on
> hard disk too - I don't know how to do that one.
Well, Red Hat use "MD5 sums" on all of the ISO images for you to
manually check if they downloaded correctly and, presuming that the
MD5 sum algorithm is working well, then the ISO images I've got should
be 100% fine...
Regards the downloading, you can _buy_ box sets of Red Hat...but you
can download it for free from the website (that "clause" in the Linux
licence about not being allowed to charge for Linux itself but only
for "reasonable costs" of media, "extras" supplied and so
forth...plus, the source code should also be made fully available and
that's in the same FTP download area...hence, to properly comply with
the GPL and Linux's version of it, then all distros _should_ be
offering some sort of similar "it's always possible to get it free, if
you don't mind the inconvenience of many hours of downloading" thing
:)...I'm going for the download option not just because it's free but
also because that actually is the most convenient way...
> The very first thing I installed was "ZipSlack" - a very minimal
distro
> that installs in a dos directory (or on a "ZipDisk", hence the
name). A
> handy way to "ease" into Linux without having to repartition
anything or
> uninstall anything. I keep it around to use as a fancy "rescue
disk" -
> it's proved useful a couple times. I suppose other distros have a
> similar scheme(?). I doubt if it'd work on NT filesystems, but if
you're
> going to be installing '98 anyway... Something like that might help
you
> figure out what's going on with the "problem" drive(s).
Well, yes...and no...because the point here is that this machine is
specifically put together as a "second machine" for running Linux
(Windows 98 may also go on there just to have a different version of
Windows for checking "compatibility" for any programs I write with the
two types of 9x / NT kernel...on the other hand, I might put "pure
DOS" on there or something else instead, as that could be handy for
testing some of the DOS code that gets posted up around here, as the
Windows "DOS box" isn't always a perfect emulation ;)...
The general idea being that I'm specifically putting together this
"second machine" as a "don't care / experiment" thing...hence, as
useful as those kinds of things are for certain conditions, that's not
really what I want here...I'm perfectly happy to re-partition
everything...in a sense, I _want_ to do so...to have proper working
"typical" versions of these OSes for "testing" purposes (hence, the
more "normal" it is with the "typical" sort of install, the better it
is for "testing" things out on ;)...
Note that I _do_ have "Cygwin" on my Windows machine already, which is
a similar thing...it's actually a novel and clever idea...there's a
"Cygwin DLL" file that provides all the Linux kernel functions and
then just "emulates" Linux under Windows (I guess the "INT 80h"
interface is out with this solution but you can always change those to
calls to the C interface equivalents :)...they've even converted
X-Windows over to a "Cygwin" version that'll run in a window under
Windows...but, though it's useful for running GCC and command-line
things, the "Cygwin" stuff is just too slow and cumbersome and
complicated to write X-Windows programs with (plus, it only has the
most basic X-Windows and no GNOME or KDE or anything)...for just
running Linux programs on Windows, it's actually a pretty clever
idea...but actually developing things for Linux this way is, in my
opinion, a bit too awkward and complicated...especially as I'm looking
to do X-Windows stuff with ASM, which really doesn't fit too well with
what Cygwin can currently manage (it's stretching itself to run
X-Windows "under emulation", as you might imagine, that trying to push
it further is too much of a "constant struggle" for it to be a good
method of development...not least because, of course, _IF_ I could
only get things to install on that "second machine", I'd have the
completely reverse in a development method that's so incredibly easy
and convenient...as I think I mentioned before, _once_ you try
cross-development like that, it's sooo easy and convenient compared to
anything else, you just can't really tolerate doing it any other
way...with two machines, for example, I can use my favourite Windows
editors and tools, listening to MP3s and so forth...then simply stick
it onto a disk, pop it into the Linux machine and run it there to test
if it works...also, if anything is taking its time on the Windows
machine, then there's always the option of playing cards on the Linux
machine while you wait...for cross-development, there simply is no
better way (although, some people may, of course, prefer to use Linux
and then merely "test" things on Windows...the other way
around...that's sort of the point of this method...it's very flexible
and versatile and simply let you use _either_ machine - even at the
same time - to get the job done...where Linux tends to do a better and
easier job of things then I can simply use the Linux box...plus, _IF_
I'd only been able to get that network card working under Windows too,
then things would have been even easier still, creating things like
"shared folders" between the two that I don't even really need to
think about transferring data across :)...
> > ... and I was coming along great with the X-Windows stuff I was
> > doing...
>
> Yeah! Been hoping for some examples from you.
And they would have turned up by now...like I wanted to provide a
simple example of a basic "hello, world!" for that poster who was
asking about X-Windows before...but, basically, until I can get my
"Linux box" to work, I can't test or properly develop the
code...that's what's most frustrating about this "install
problem"...it's just totally ground all my plans to a halt until I can
work out what's wrong to fix it...
> I've seen a couple
> examples from Numi_tor - one using xlib directly and one using the
Gnome
> Toolkit - they both look pretty complicated - even more
"housekeeping"
> than Windows to do before you can do anything.
Odd; With XLib directly and the most _minimal_ "hello, world!", I
always liked X-Windows because it had _less_ nonsense involved and far
more flexibility...again, I'd demonstrate this with some code but,
without the Linux machine, I can't really test it (plus, though I'm so
used to writing Windows code that I can put together a "Hello, world!"
blind, so to speak...I've done less X-Windows (and Linux too) stuff
that I'd still like to have access to "man" and "info" pages and
references too :)...
But, anyway, talking about it instead of example code, I found
X-Windows to be _better_ on a number of scores...at least, though it
may not always look as pretty or be as responsive of Windows, the
underlying X-Windows design has a better style to it...Windows "wins"
on "fancy graphics" because Microsoft know the old trick of "as long
as it looks good on the screen, most users are fooled by this into
thinking its a better product...when, in fact, 'under the hood' it can
be a hundred times worse, as long as it looks a hundred times _better_
on the screen, it'll surely sell like hot cakes"...hence, X-Windows is
a bit complicated to understand when you first read the documentation
because it's going on about "bitplanes" and stuff, where GDI doesn't
mention these...but, of course, what has to be remembered about
X-Windows is that it works on a _wide, wide range_ of computers and,
subsequently, hardware that this stuff does make plenty of sense
("bitplanes" were _THE_ standard way of doing graphics, right up until
"mode 13h" on the PC introduced the simplicity of "one byte per pixel"
and DOOM was written for it and it changed how things worked from then
on..._every_ machine - including PCs - were "bitplaned" right up until
this "cheap hack" for getting a 256 colour mode on hardware that only
had 4 bitplanes (16 colours) hard-wired from the EGA days...as is
quite typical, this industry has been ruled by "happy coincidences"
and "accidental empires" all over the place...because, in fact, when
you had a low (16, 8, 4 colour displays :) amount of colours,
bitplanes can actually be superior...more complicated but, as can be
done in VGA mode 12h for example, multiple pixels all written at the
same time...in fact, much like "video killed the radio star", DOOM
forever killed bitplanes...not only did it use mode 13h's then-unique
method to very good effect, it made the 3D textured polygon
"king"...and bitplanes _aren't_ at all useful when you have a high
number of colours with every pixel in a scanline a different colour
(typical of texture mapped polygons :)...actually, it was Wolfstein
that first picked up the knife to kill bitplanes but DOOM knocked
everyone's socks off in such a permanent way that that was the day
they truly died...now merely a "historical curiosity"...only bothered
to be used for VGA's "mode 12h" for some sort of "safe mode" with
reasonable resolution (because beyond mode 12h, you have to start
going VESA VBE and device drivers and DirectX / OpenGL...the entire
landscape changes once you step beyond 640 x 480 x 16 colours ;)...
Anyway, this "bitplanes" thing can certainly look confusing and put
people off X-Windows because Windows does look at graphics in a more
simplified way (but, then again, that's a trade-off...if you know what
you're doing, then "doing it yourself" very often gets better
results...Windows has simple "generic" libraries to do the hard work
and all the usual arguments of "DIY versus libraries"
apply...although, with "accelerators", Windows wins an edge by using
that hardware capacity to "patch over" any "overhead" the library
stuff may bring...meaning that Windows _does_ tend to do better but
only because the _hardware_ is doing all the "grunt work"...a
similarly wised-up "accelerator-aware" set of X-Windows API would
catch-up and theoretically exceed because of the more "low-level"
perspective...X-Windows is only really "behind" on such things because
it's designed to work on _any_ hardware - and _does_ do so - and has
to take a "generic" perspective in its API...it may be hard to
appreciate when you don't _see_ X-Windows perform better on the screen
because Windows is exploiting all the PC hardware, but X-Windows
really does - on many scores - have a much better underlying design to
it :)...
Actually, X-Windows _lacks_ the "overhead" of Windows almost
completely...the steps necessary to get things going are much more
logical and don't have some of Windows' weird restrictions...
Plus, you have to realise the difference in "attitude" involved...for
instance, with X-Windows, you're program needs to "connect to the X
Server"...something not done in Windows...but, then, the X-Windows
solution is a whole lot better...what makes something an "X-Windows
application" is the fact that it connects to an X server...under
Windows, it's because the "GUI" bit is enabled in the
headers...Windows is "hard-wired" to do local applications running on
a single machine and you have to "jump through hoops" for any "remote"
operations...while X-Windows is "client / server" from the ground
up...so, yes, you have to "connect to an X server" but, then again,
this also means you can connect to an X server on the other side of
the planet over a broadband link...it means that you can connect to
_more than one_ server at the same time (one program controlling the
output on multiple machines :)...and so on and so forth..."remote"
applications simply require connecting to a "remote" server...and then
that's it...otherwise, the rest of the application code is
_identical_...you do not need to make any "allowances" because
X-Windows does those for you (when it sees that the server is
"remote", it automatically handles passing the stuff over the
connection between client and server :)...
Also, there's other points of interest...in Windows, you either have a
"console" or "GUI" application...if you choose "GUI" then even if it's
started from a command prompt, the process doesn't inherit the
"standard handles" (this is a very annoying thing I've encountered,
trying to create "dual mode" programs...programs that act as
command-line utilities if given command-lines but when given nothing
on the command-line, they instead open up a GUI interface...so that
you can use it "batch" or "interactive" as is most appropriate
:)...again, X-Windows isn't really "thinking" like this...ordinary
command-line applications and X-Windows applications are not any
different...the _difference_ comes about exclusively because you make
a call to "open a connection to an X server" (or, as noted, to _more
than one_ X server for some really mad distributed functionality
;)...you link your programs with XLib simply to get the function that
opens the connection (if you knew the exact details of the "X
protocol" then you can theoretically by-pass this and do it
yourself...but that's "non-portable" - which is more of a concern on
X-Windows because it really _does_ run on PCs, Sparcs, Silicon
Graphics workstations and so forth :)...
Otherwise, there's no "window class" nonsense...nor is there "window
procedure" nonsense...once connected to the X server, you need a
"message queue" (but that's somewhat non-optional, if you think about
it, without a "message queue" how exactly are you expecting to receive
messages? ;) and then you just pop those messages off and react to
them...when you want a window, you simply "CreateWindow" without any
nonsense "window class" beforehand...that's how it _should_ work...
If you want "window procedures" then simply create an ordinary
procedure that accepts information about a message, containing one of
those big "switch" statements...and then jump to that procedure with
the new messages...on the other hand, you can deal with the messages
all inside "main", if you like...or perhaps have "common" procedures
between different types of windows...which is, of course, how you'd
also get any "window class" functionality...if you want a bunch of
windows to all be of the same "class" then, sure, write a "window
procedure" and then pass those same messages to the same procedure
(which'll react in the same way to those messages, making them all of
the "same class" :)...
On the other hand, you can structure them completely differently, if
you like...this isn't "overhead" and "restriction" in the same way as
Windows' "window class" and "window procedure" nonsense is...this is
simply _choice_ because X-Windows provides the basic services and then
it's up to the program to format these as is most appropriate to the
task in hand...
Plus, X-Windows has a more sensible attitude to "messages"...under
Windows, every application gets every message...even if your program
only processes WM_CREATE, WM_PAINT and WM_DESTROY, it still gets a
stream of WM_MOUSEMOVE, WM_SETCURSOR and so forth...what does it do
with these other messages? Why, it wastes time simply sending them
"return to sender" with "DefWindowProc"...if you stop and think about
this, then a simple easing up on the "control freak" attitude produces
a far more sensible approach...the _application_ knows what messages
it processes, right? If you send it anything else, then it's just
"return to sender" that it was slightly pointless sending it in the
first place...
Well, though X-Windows isn't perfect by any means (only includes /
excludes "ranges" of messages rather than individual ones), it at
least allows an application to simply say "look, I don't ever read the
mouse so there's totally no point sending mouse messages to this
window"...of course, X-Windows has the right attitude here because all
these messages _may_ be being sent over a 56K modem connection to the
other side of the planet...it can't really afford to be sending
pointless messages for no good reason...hence, there is a means to
simply specify "don't need those kinds of messages, don't even bother
sending them"...
If someone out there is considering making their own GUI system, then
I'd strongly advise that you take X-Windows attitude as a "role model"
here rather than Windows...in fact, take it to the extreme once more,
I'd suggest...default to sending _NO_ messages (except, perhaps,
"obvious" ones like "window creation", "window destroy", etc. ;)
_UNLESS_ the application specifically requests it...and the perfect
opportunity there is to have the application simply specify what
messages it actually processes while creating a message queue...but,
also, include some "AcceptMessages" API where you can turn messages -
individual or "ranges", by having a "start" and "end" parameter which
can simply be the same message value for individual messages - "on"
and "off"...there's the possibility that a window _may_ initially
check the mouse location once or twice and then never bothers ever
again and so forth...
Windows spends an awful lot of time sending pointless messages...this
change of style alone will win any GUI masses of extra processing
time...there's also the possibility to "propogate" this benefit
downwards...that is, a window doesn't accept mouse messages...well,
not only does the GUI not need to bother that window with mouse
messages but the OS can even inform the mouse driver not to bother the
OS when it's in a certain area of the screen...the "pointless
messages" can be propogated backwards to _stop_ as soon as
possible...apply this throughout the system from the ground up - plus,
Windows wastes thousands of cycles in simple "overhead" handling a
message, which you could work to improve upon - and you can "release"
so much extra CPU power into a system that's far from
insignificant...ever seen the red "kernel time" on Windows' task
manager's graph of where the CPU power is going? Ideally, if you were
to have a similar graph in some OS you wrote, then the thing to aim
for is that the red graph is hardly there at all...it would be
impossible to make it completely vanish but that, in a sense, is the
"ideal" you should be working towards...an OS is _BAD_ and _WRONG_
when it's using CPU time...necessarily, it does have to use some in
order to keep the system going...but the system is the
_applications_ - the OS merely there to "accomodate" them - and they
should effectively totally rule the system as far as is
possible...that understanding and attitude is _severely missing_ from
Microsoft and is one place where some other GUI / OS can _do things
properly_ so as to begin to blow Windows out of the water...because,
yes, that's the thing that is also missing...people see Goliath and
truly believe he's undefeatable...this is something that I don't
believe at all...now, just as with the fabled Goliath, you don't stand
a chance in a straight-out battle (the soldiers before David attempted
that and failed miserably ;)...but, whether the "Art of War" actually
says it like this or not, the best warrior wins without fighting (or
by doing the least amount of fighting necessary)...
David, of course, was _guaranteed_ to win...now, that may initially
sound odd but if you look at the story, it was actually
_inevitable_...this is the underlying message in the story that's
worth making obvious and clear...the reason being that the soldiers
before David walked up to Goliath - within Goliath's reach - and tried
to simply "out-strength" a giant when, clearly, he has more strength
than ten of these soldiers put together...but note that David's
approach was _different_ and _that_ was why he defeated Goliath...he
used a sling...a "missile weapon" that could cause damage from afar
within ever putting David in danger...just a case of David slinging
stones until one of them hits that "right spot" that brings the giant
down...many people understand the "even the smallest can defeat the
biggest" part of this tale but many overlook the actual reason _why_
David was sure to win...though his attacks may not be as strong as
Goliath's, Goliath was unable to strike so all that strength he had
amounted to _absolutely nothing_ because he could never use
it...however "small and weak" David's attacks in comparison, he was
_able_ to make his strikes...eventually, he would find that "weak
spot" which takes him down in one blow...or simply pelt so many stones
at Goliath that Goliath gives up or gets slowly and surely stoned to
death...
Feel free to be impressed by Goliath's size but, if he's "a bit
thick" - all brawn and no brains - then fear is completely
inappropriate...and - *ahem* - "640KB is enough for anyone"? Need we
say any more? ;)
> I haven't done anything
> with Xwindows yet - still haven't got "console mode" figured out...
It's not particularly difficult...especially if you've gotten the
rough idea of what all that "event-driven" nonsense is from writing
some Windows code...in essence, it's just a program that waits for
messages and reacts to them as they come...
And, in fact, in X-Windows' case, it isn't as weird as Windows in
insisting that everything is in "window procedures" and so forth that
it's _easier_ to make the transition with X-Windows...other than
linking to the XLib library (just so that you're linking to the API
functions :), then it's no different - except for the functions it
calls and the "wait for messages" structure - to other Linux
applications...
> > ...and doing it all on Linux
> > would be a pain because all my best development tools are on the
> > Windows machine...
>
> Curiously, the Unix-folk claim that *their* tools are better - and
have
> invented Cygwin so they won't have to leave 'em behind when
developing
> for Windows! The famously user-friendly Unix interface bolted onto
the
> lean, mean, stable Windows chassis. A dream environment! (a
nightmare's
> a dream, innit?)
Oh, don't get me wrong...UNIX tools _are_ better in some
regards...that is, "GREP" is one of those indispensable UNIX
tools...they are able to do some pretty impressive stuff from a simple
command-line on occasion...and there's just tons of these "short and
sweet" utilities (having "many small" which you jigsaw together to do
your task, is usually far more useful and flexible than "one big" -
which tends to be the Windows way - in general :)...
> What would you miss most that you can't have in Linux?
My MP3 collection, my rather large "tech-docs" folder where I store up
all my references (makes no impact to my many tens of GB hard disk on
the Windows machine, would be rather large though for the smaller
"second machine"), Windows Solitaire and Spider Solitaire games, the
text editor that I've gotten so used to using that I can do things
really quickly with it, my "bells and whistles" graphic editor program
(professional quality and wasn't cheap to get ;), etc....
Yes, nothing _technical_...that's the grand mistake "geeks" always
make..."technical" and "techno-fix" is often the least important
thing...Windows and the stuff I have on the machine are "comfortable"
and I'm used to them...and, therefore, I can develop happily and
quickly...
Oh, trust me...I've used Cygwin to run Linux GCC on Windows in order
to cross-develop for a PDA, for example...I've cross-developed onto
circuit boards that have no "monitor" or anything (switching on and
off small LEDs on the front panel to let me know things were working
properly ;)...the other day, I was bored round a friend's house and
they had a PC but absolutely NO development tools at all and I decided
to waste time - had to wait around for hours one day - waiting by
challenging myself to a simple but tricky task: "using a command
prompt and debug, slowly build a hex tool, then an assembler then work
up to writing some simple DOS-based .COM game" and the trick being
that I had NO references, NO tools or anything...only what comes "out
of the box" - the command prompt commands, DEBUG, etc. - with any
Windows installation (never actually completed this task because it
takes longer to do than I actually had to wait...but it was a way to
waste time while I was waiting...and I had the plan in my head that it
was completely possible...it just would probably take _days_ to
complete on the first attempt...using DEBUG itself in order to get the
opcodes for the assembler...only those DOS / BIOS interrupts - like
"mov ah, 00h; int 16h" for keyboard input - that I could actually
remember off the top of my head :)...also, I started out programming
on those little 8-bitters where 64KB was "elephant memory"...
I _can_ work, if necessary, in some bloody awful "conditions"...but,
having two machines - one running Windows (which I've built up into a
nice, comfortable system :) and one running Linux - is the perfect
solution...note that when Linux _is_ the better choice, I simply use
the other machine for that task...but, well, I use Windows primarily
and it's my "main machine"...plus, there's the "little things"...like,
say, Randy's PDF documentation...reading PDFs is easy in Windows, I've
got the plug-ins already but it would require a lot of downloading and
so forth to get a Linux installation up to the same levels of
"convenience"..._IF_ I only had Linux then that's _exactly_ what I'd
do...and, over time, if I add to the Linux system then I could perhaps
get it up to a level where it doesn't particularly matter which I
use...and cross-developing, I _AM_ using Linux and an "X-term"
completely for the actual compiling and testing and stuff...but, well,
though "gedit" is more or less "Notepad which can handle multiple
files at the same time", it's just not as convenient or am I
comfortable using it as the IDE that I've had for _years_ that I know
it backwards...
This is the _really weird_ thing in the Linux world...they are
constantly on the defensive and so forth...you know, if someone isn't
singing their praises that Linux is "perfect" in _ALL_ possible
situations, then I'm "evil"...I _like_ the bloody thing...that's why I
am - if people are actually rememebering _what_ this is all about -
developing X-Windows applications for Linux...how I'm Hoping to help
Randy get X-Windows properly into HLA that doing GUI applications is
_exactly_ as "easy" and "without modification" that HLA can already do
for "console" applications...that, in fact, is _exactly_ an exercise
in leveling the field between the two so that, yup, it _doesn't_
matter which you use...
I like that KDE thing (the best yet is that you can get it to put a
menubar over the top of the screen like an Apple Mac...I've always
liked that style - certainly is more welcoming and useful to "newbies"
to be able to _read_ the available commands off a menu - and KDE is
"politically correct" enough in its options that you can get it to act
like any GUI...and I was using plenty of GUIs _before_ Windows, that
I'm NOT always impressed with the "Windows way" and _DO KNOW_ that
there are better ways...Windows' "taskbar" and "resizing borders" is
the better on that score, Apple's "menu at the top" is better on other
scores...and I totally _prefer_ a desktop where the drives are listed
as icons rather than some strange abstract "My Computer" (which is a
slightly presumptious name, I've always thought...what if it's a
machine in a school or business or at a friend's house? Then it's NOT
"My Computer" at all...although, it can be excused because you know
what it's trying to say - being "colloquial" in the name delibrately
to make it sound "friendlier" - it's a bit presumptious and can,
therefore, often be simply factually incorrect ;)...
This ain't in any way some "anti-Linux" attack...far from it...just
that when things aren't right, I always speak my mind...things aren't
100% right in all departments...there will _NEVER_ ever be such a
thing in existence on this Earth that fits that description, as
"NOTHING is perfect"...I think the "issue" with Linux is that people
actually take comments about Linux _personally_...you know, I say "oh,
I can't stand that 'vi' editor" and it's like I've just called their
mother "ugly" or something...look, _I_ can't stand it...doesn't mean
that _other people_ can't like it and think it's the best thing since
sliced bread...it's just the way things are...no need to take it as a
personal assault on all of creation or anything...chill out, Linux
people...if anything, I totally "come in peace" here to help
out...this whole "Spanish Inquistion" is _part_ of all the "voodoo"
nonsense that I was actually pointing out...you just don't expect such
things when you're just trying to make honest comments and ask simple
questions and make some jokes to Hopefully bring a smile to someone's
face somewhere :)
> > I'm also wondering whether to write a book on the psychology of
how
> > people describe their machines...
>
> Mac users get to say, "Mine's blue!" Well, I suppose we could
> spray-paint 'em...
Ooh, those iMacs are works of art...actually, literally, because they
won design awards that covered a whole bunch of stuff that it really
did have stiff competition that winning with something so usually "not
even considered" as the design of a computer _is_ earnt and
deserved...I even saw a bath in a magazine which actually was
"inspired" by the iMac design, having that same "semi-transparent
blue" plastic bit thingy...so you could colour co-ordinate yourself
while relaxing in the bath using your iMac (though, hmmm, encouraging
the use of electronic equipment in the bathroom? Someone's not being
totally "safety conscious" there, eh? ;)...much like the Playstation's
joypad - which also won awards for brilliant design - I've had to
agree with the judge's decisions...these two things really _were_
fantastic design achievements...really cool-looking but still
perfectly practical and functional...
Mind you, have you seen the "explosion" of PCs with moulded plastic
blue bits and odd curved shapes, trying to copy the iMac? Slightly
pathetic, really, as none of them come close...they are only thinking
"jump on the bandwagon", they aren't thinking "art" (but also
"functional"...because the hard drive and stuff is actually also
_inside_ that monitor, meaning that the entire system is nice and easy
to carry around too...as it's all in one place and only needs the
keyboard and mouse attached...though, with wireless stuff, even that
won't be necessary ;) which is what made the original design so
cool...
I'm not quite so impressed with the "lampstand" design...it doesn't
quite look as "cool-looking" as the original iMac designs (which, by
the way, you can get in a silly amount of different colours...there's
even "zebra stripes" and "dalmation spots" and mad-looking ones like
that :)...but it's still innovative and impressive design...just the
original iMac designs were so brilliant, even something great looks a
bit crap in comparison to it...
When I went into the store to buy my current PC, there were all these
"bad iMac impersonation" PCs...you know, strange curved bits of
coloured plastic for no particular reason but to try to pretend that
the PC was actually an iMac...and when I ordered my PC, he asked what
sort of box I wanted it put into (because it was one of those "put
together to order" things :) and I looked at all the boxes...and the
way I put it was: "_IF_ there's one here that looks as cool as an iMac
then I'll take it...but if it's not quite that good, then go to the
total other extreme and give me a plain boring beige-coloured
box"...you know, it's either drop dead gorgeous looking or, if that
can't be managed, then just don't bother with "looks" at all and give
me something boring, plain and simple...in short, I said "I either
want the real deal or don't bother"...the box I have, by the way, is
plain with only a few small bits of blue plastic on it (which are all
removable :)...I'm pleased to say that the guy behind the counter
_totally_ understood what I meant and made the right choice...you
know, either you drive a drop dead gorgeous Porsche / Ferrari or you
drive a "functional but ugly" Skoda or whatever...the real deal or
nothing at all...so, to sum it up in one sentence: "I have no time for
Toyota" ;)
Anyway, what's wrong with beige? I mean, it's the perfect "camouflage"
colour for hiding vomit stains...very "functional"...hehehe ;)
Beth :)
- Next message: Beth: "Re: OT: my new PC rocks!!"
- Previous message: Madhur: "why MASM generated this code"
- In reply to: Frank Kotler: "Re: OT: my new PC rocks!!"
- Next in thread: pH: "Re: OT: my new PC rocks!!"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
