Roy Schestowitz wrote in <1492620.uhl6PK4TOC@xxxxxxxxxxxxxxx> on Mon May 1
2006 12:57:
>
> Over here, it was at most 33 MBit/Sec for Windows and 100 MBit/sec
> (maximal speed) under Linux. The system was idle enough to permit other
> programs to run and remain fairly responsive (especially the JRE 'stuff',
> for whatever reason). Call it "stress testing" if you like, but it comes
> to show how slow and cumbersome Windows has become. It is _not_ suitable
> for brute-force, 24/7 work (with or without the presence of the
> user/admin).
>
Oh, what I'd give to have a pipeline like that. The bottleneck wouldn't
likely be my connection at that point, lol.
>
> If it takes 10 minutes to restore a desktop session with all the needed
> connections and precesses (I have well over 100 of them), I cannot afford
> fragility. Fragility, as I have experienced in my cousin's house (Windows
> XP) costs too much time, effort, and ultimately patience.
>
Totally agreed. My session only takes about sixty seconds to come up, but I
have an awful lot running in it that is preserved across boots and the
like. I occasionally run into a hiccup here or there, but never anything
that's cost me data or anything more then 30 seconds of my time to bounce a
process. Things have gotten *worlds* better then they used to be, say, in
the early-to-mid 90s.
>
> I never wish to spend any time on maintenance either. I don't enjoy it.
> Nobody does, but others treat it as a necessity and take it for granted. I
> want things to be completed as soon as possible and require the least user
> intervention. GNU/Linux gives me the ability to come in day after day,
> punch in password to unlock, and always resume without taking time aside
> to update, to defrag, and to worry about viruses or malware, let alone the
> integrity of my data.
>
I wouldn't say that my Linux boxes are exactly maintenance free -- though I
have it all pretty much automated for the most part. They can even page me
if they're in distress (a backup fails, or something else bad happens).
>
> Data (information) is the bread and butter of all who go beyond basic Web
> surfing and Webmail. GNU/Linux and its origins have concentrated on this
> aspect for decades, in the enterprise and primarily its datacentres. Thus,
> they all handle it so well, whereas Windows has been progress-driven
> (read: more bells-and-whistles), concentrating on UI and ease of use (by
> an outsider as well, if you know what I mean).
>
Yep. Though I have to wonder why Windows hasn't taken a data-centric
approach. Users should be more familiar with their data then they are, I
think. I liken it to the person that calls me up saying that they're out
of space on a drive and they want to know what they should remove to free
up space. I tell them, that they should know what they've put on the
computer, so they should know what to remove, and they're generally
confused. In the case of someone who *isn't* running Windows, I can
say, "well, you have X, Y, and Z -- which is it that you don't need
anymore," without even leaving my house. They call me on the phone, and
I'm able to ssh into their computer (provided they're paying me for
maintenance) and fix problems remotely. Sometimes, it's as simple as the
user didn't realize that they'd be using so much storage and so I can divy
up the storage right from the comfort of my chair, and by the time I hang
up with the person, they're back in business.
>
> Dvorak's crazy article (nothing out of the ordinary) on Open Source Vista
> should be folded, squashed, and thrown out the window. 60% of the code in
> Vista is awaiting a re-write, according to Microsoft. It is too early to
> expose it. It is not ready for prime time, according to the EU (Microsoft
> offered to sell the code to selected developers).
>
I think that they'd benefit from it -- but not just because of opening it
up, more because the people that would come in and start working with it
would be more quick to scrap things and rewrite them, since they're not
exactly married to the code.
>
> You see, Michael, when they write in a closed-source context, the quality
> of code developed is affected accordingly. It is not visible. It needn't
> necessarily be tractable (think of legibility of todo notes that you use
> and bin a few hours later). I believe I know this rather well because I
> made all my code public. I also see code of others whose work I'm a part
> of -- either as a contributer or a user (I often need to hack/tailor
> projects at source level).
>
Totally agreed. Unfortunately, Windows isn't something that we can look at
and go, "You know what, it does this pretty badly, let's just go ahead and
fix it here..." without going, "Well, function 'x' in DLL 'y' doesn't work,
so in order to fix it, we have to re-write the entire DLL and introduce
probably more bugs then we're fixing..." and that probably even violates
something in the EULA of Microsoft's -- not terribly sure about that,
though.
>
> All in all, forget about Open Source Windows for the time being. Internet
> Explorer has got its own Bugzilla copycat recently, but noone can submit
> patches. The code remains closed source.
>
Yeah. Though, people have been cracking source for a long time, and
creating binary patches to remove or add specific functionality. If IE
were worth it, I'd suspect that a community like that would've sprouted by
now.
I actually have to wonder -- is there any country that isn't covered by
Microsoft, or that doesn't enforce Microsoft's copyright, wherein a
developer could legally reverse-engineer Windows? If so, why hasn't
something sprouted from that direction? It's not like Microsoft could stop
something like that from working against them...
>
> I am probably too young to be familiar with ZipSlack. I had to look it
> up...
>
> http://en.wikipedia.org/wiki/ZipSlack
>
Eh, I don't know. IIRC, they're still making ZipSlack with new Slackware
releases. I used it *once* when I needed to take Linux with me somewhere
and LiveCDs weren't that popular yet because of the lack of El Torito in
systems. Now that computers can boot from USB, and El Torito is generally
supported throughout the PC industry (and even if not, it's available in a
boot manager that can be saved on a floppy disk), ZipSlack has kind of lost
its appeal. It's quite trivial to carry around a USB flash drive or an
ATAPI drive in a USB-connected casing and be able to have real Linux
filesystems instead of running with UMSDOS. :-P
>
> Earlier on I forgot to mention people whom I heard migrated to Linux owing
> to major data losses in Windows. I am very serious about this and I know
> how devastated I would be if I ever lost data. I sometimes do a filecount.
> I still mirror my hard-drive in 3-4 places due to fear of physical errors
> (bad sectors).
>
I've seen too much data lost that way, myself. I mentioned somewhere else
in here recently about how I've seen DOS / Windows 3.x / Windows 9x/Me
trash my data like nothing else ever has before. I've used non-journaled
filesystems on Linux for a long time before ext3 and such, and never had
any problems with data lossage, though I've had a hell of a time with it on
FAT-based filesystems (even with Linux). FAT just isn't designed all that
great -- it was definately not designed with disk buffers and caching in
mind, and hell, the very first version of DOS didn't even implement
directories in its version of FAT! (See
http://en.wikipedia.org/wiki/File_Allocation_Table#FAT12 for some
background on that, if you don't recall it or wasn't there for it... I
wasn't actually there for it, but when I started using computers, MS-DOS
2.11 was what came with the computer, and MS-DOS 3.something was out at the
time...)
I never did lose much data, until Windows entered the scene. I had less
data eaten by bad 5.25" floppies then Windows has consumed in my lifetime,
of just my data alone.
- Mike
|
|
