Linked by Howard Fosdick on Mon 6th Sep 2010 21:56 UTC
Windows In previous OS News articles, I've claimed that mature computers up to ten years old can be refurbished and made useful. My last article identified and evaluated different ways to refurbish these computers. One approach is to keep the existing Windows install and clean it up. This has the advantage of retaining the Windows license and software, the installed applications, and the existing drivers. But it takes some work. In this article we'll see what this entails.
Thread beginning with comment 439635
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[6]: Nice article.
by pgeorgi on Tue 7th Sep 2010 09:30 UTC in reply to "RE[5]: Nice article."
pgeorgi
Member since:
2010-02-18

I only hope and dream that maybe Microsoft will put its foot down and make the changes even in the face of much protest from the unwashed masses.

They did, to a degree: UAC. Notice how everyone whined about it?
It's a good measure to tell people (and developers, directly or by proxy) that there's something wrong with their apps, so there's pressure on devs to minimize the problems that lead to UAC notices (and they did), while not breaking ancient, unmaintained legacy apps (they just get a tad annoying, hopefully pushing users to plan to migrate off of them eventually).
Windows 8 or 9 might do away with UAC, _finally_ breaking those ancient apps (or transparently pushing them in a sandbox - that already starts in Win7 with the namespace virtualization), while giving everyone a chance to fix things in the meantime.

That's what I like with Windows or Solaris: Their maintainers care about compatibility, while planning how to move forward with hacks like these to push people in the right direction. (Sadly on Solaris it's less so since they started with OpenSolaris)

On Linux, you simply get changes thrown at you, forcing you to cope _immediately_ with them (or lack new features because to update, you'd have to update libfoo, which requires udev no older than x.y, which requires you to switch the device detection mechanism, which ... and so on) - that model is good for 0.0.x versions, where experimentation happens, but I really despise it for "mature" systems (such as those I'd like to work with daily)

Reply Parent Score: 2

RE[7]: Nice article.
by kaiwai on Tue 7th Sep 2010 10:55 in reply to "RE[6]: Nice article."
kaiwai Member since:
2005-07-06

They did, to a degree: UAC. Notice how everyone whined about it?
It's a good measure to tell people (and developers, directly or by proxy) that there's something wrong with their apps, so there's pressure on devs to minimize the problems that lead to UAC notices (and they did), while not breaking ancient, unmaintained legacy apps (they just get a tad annoying, hopefully pushing users to plan to migrate off of them eventually).
Windows 8 or 9 might do away with UAC, _finally_ breaking those ancient apps (or transparently pushing them in a sandbox - that already starts in Win7 with the namespace virtualization), while giving everyone a chance to fix things in the meantime.


But the poor communication explaining UAC to the average user didn't help either; if the average user knew that the UAC could be avoided if the software vendor actually updated their software then you might see the end user putting the hard word on software by pestering them.

That's what I like with Windows or Solaris: Their maintainers care about compatibility, while planning how to move forward with hacks like these to push people in the right direction. (Sadly on Solaris it's less so since they started with OpenSolaris).


True, but even with Apple they're pretty fair with their transition; the only things I've seen broken on movement between different versions of Mac OS X are vendors using private frameworks that should never have been used in the first place.

On Linux, you simply get changes thrown at you, forcing you to cope _immediately_ with them (or lack new features because to update, you'd have to update libfoo, which requires udev no older than x.y, which requires you to switch the device detection mechanism, which ... and so on) - that model is good for 0.0.x versions, where experimentation happens, but I really despise it for "mature" systems (such as those I'd like to work with daily)


Agreed; and worse comes when there is no smooth transition from one to the other; you can put in the older way of doing things but then a whole heap of interoperability problems rear their ugly head.

Edited 2010-09-07 10:56 UTC

Reply Parent Score: 1

RE[7]: Nice article.
by lemur2 on Tue 7th Sep 2010 11:40 in reply to "RE[6]: Nice article."
lemur2 Member since:
2007-02-17

On Linux, you simply get changes thrown at you, forcing you to cope _immediately_ with them (or lack new features because to update, you'd have to update libfoo, which requires udev no older than x.y, which requires you to switch the device detection mechanism, which ... and so on) - that model is good for 0.0.x versions, where experimentation happens, but I really despise it for "mature" systems (such as those I'd like to work with daily)


Solution: Install a LTS (Long Term Support) Linux distribution with a back-ports repository.

Here are two candidates:
http://distrowatch.com/?newsid=06030

http://distrowatch.com/?newsid=05334

You will not then have to contend with any of the problems you claim, yet the installations will be supported (with security updates but not necessarily feature updates) for a long time into the future (at no cost other than a bit of Internet bandwidth).

PS: On Linux, you do not have to install updates immediately, or indeed at all if you do not want to.

http://ubuntuforums.org/showthread.php?t=541173
Is there a way I can somehow "blacklist" udev so that running apt-get dist-upgrade will upgrade every package except udev?

...

Highlight the package in Synaptic, then go to "Package" and check "Lock Version".


Edited 2010-09-07 11:44 UTC

Reply Parent Score: 2