| 1 |
On Wed, Feb 1, 2012 at 12:36 AM, Dale <rdalek1967@×××××.com> wrote: |
| 2 |
> J. Roeleveld wrote: |
| 3 |
>> |
| 4 |
>> On Tue, January 31, 2012 6:30 pm, Walter Dnes wrote: |
| 5 |
>>> On Tue, Jan 31, 2012 at 06:05:12PM +0100, Michael Hampicke wrote |
| 6 |
>>>>> Sweet. I had 15 minutes in the office "how long before someone makes a |
| 7 |
>>>> pointless, unrelated Windows insult out of my post" pool; I just won |
| 8 |
>>>> $5. |
| 9 |
>>>> |
| 10 |
>>>> I was using Win3.1 - and was happy with it |
| 11 |
>>>> I was using Win95 - and was happy with it |
| 12 |
>>>> I was using WinNT4 - and was happy with it |
| 13 |
>>>> I was using Win2000 - and was happy with it |
| 14 |
>>>> I was using Win Server 2003 - and was happy with it |
| 15 |
>>>> I was using Win7 - and was happy with it |
| 16 |
>>>> |
| 17 |
>>>> And I am also a Linux SuSe user since 6.0 and Gentoo user since |
| 18 |
>>>> 1.something (but up until now just on the servers). |
| 19 |
>>>> |
| 20 |
>>>> I made the final switch from Windows to Linux on my Workstation (Gentoo) |
| 21 |
>>>> and Notebook (Lubuntu) only a few month ago. |
| 22 |
>>>> |
| 23 |
>>>> So please, don't accuse me of making Windows insults. |
| 24 |
>>> |
| 25 |
>>> I feel that Win98SE was the best Windows ever, and could've been even |
| 26 |
>>> more of a killer if Microsoft hadn't so stupidly tried to ram ActiveX |
| 27 |
>>> down people's throats. Remove ActiveX, and 99% of "drive-by-downloads" |
| 28 |
>>> would've disappeared. WinME was a sad joke, however. |
| 29 |
>> |
| 30 |
>> I enjoyed MS Dos, then played a bit with MS Win3.11, MS Win95 and MS Win98SE. |
| 31 |
>> However, for important stuff, like day-to-day desktop, I switched to Linux |
| 32 |
>> in 1997. That was the last time I lost files due to a crash of MS |
| 33 |
>> Windows... |
| 34 |
>> |
| 35 |
>> -- |
| 36 |
>> Joost |
| 37 |
>> |
| 38 |
>> |
| 39 |
>> |
| 40 |
> |
| 41 |
> |
| 42 |
> When 3.1 came out, I changed jobs. Swapping 15 floppies is no fun to |
| 43 |
> me. Funny, reinstalling fixed the problems back then and it still is |
| 44 |
> the best way to fix windoze. |
| 45 |
> |
| 46 |
> < sighs > |
| 47 |
|
| 48 |
Actually, the reason for that's pretty easy to explain. It's because |
| 49 |
Windows, unlike every major Linux distribution since Apt, wasn't |
| 50 |
designed around pulling software from centralized repositories. |
| 51 |
Instead, ISVs were expected to provide installers, which users were |
| 52 |
expected to obtain from outside channels and run. That seems archaic |
| 53 |
to Linux users, but even Red Hat was like that before yum. |
| 54 |
|
| 55 |
Since there was no centralized, curated software repository maintained |
| 56 |
by people ensuring things worked properly together, you got everything |
| 57 |
from DLL hell to developers violating Microsoft's recommendations |
| 58 |
(and, considering that Microsoft *designed the platform*, you can |
| 59 |
consider their recommendations as part of the platform spec) and good |
| 60 |
development practice. So you have things like: |
| 61 |
|
| 62 |
* People bypassing APIs and munging registry keys directly. This would |
| 63 |
be like a Linux app going in and modifying Debian's package database |
| 64 |
without going through an intermediate library kept in lockstep with |
| 65 |
the package manager code. Eventually, one's going to behave in a way |
| 66 |
the other isn't going to expect, and either the package database will |
| 67 |
become corrupt ("f'ing $OSVENDOR! Their stuff keeps breaking!", the |
| 68 |
user will curse), or the application will stop working ("F'ing |
| 69 |
$OSVENDOR! They keep breaking my stuff!") |
| 70 |
|
| 71 |
* People not bothering to understand DLL search paths, and getting |
| 72 |
into the habit of dropping their DLL into the SYSTEM32 folder. That |
| 73 |
would be like manually building and installing a package to /usr/ |
| 74 |
instead of /usr/local, or a library in /usr/lib or /usr/local/lib with |
| 75 |
an improper soname. Eventually, you risk changing the behavior of an |
| 76 |
unrelated app, or having an unrelated app change your app's behavior, |
| 77 |
all because a couple DLLs had the same name and no differentiating |
| 78 |
metadata. |
| 79 |
|
| 80 |
* People only ever testing their programs while they have |
| 81 |
Administrator privileges, and so their programs only ever work |
| 82 |
correctly while running as Administrator. This would be like an app |
| 83 |
found in /usr/bin assuming it can write anywhere it pleases, call any |
| 84 |
API call it needs, and doing some marginally unsafe things with system |
| 85 |
calls. To get it to work properly, you'd have to make it suid root, |
| 86 |
and it'd be a vulnerability vector. |
| 87 |
|
| 88 |
The analogies aren't perfect, but the points still stand. Sad thing |
| 89 |
is, if and when Microsoft takes steps toward a repository model (these |
| 90 |
days, people like to call them app stores) they'll be lambasted as |
| 91 |
being evil for applying a gateway to the platform, even though it's |
| 92 |
going to be a necessary step to fixing a lot of what's wrong with the |
| 93 |
development culture on that platform. |
| 94 |
|
| 95 |
Linux isn't perfect in these regards, but the combination of being |
| 96 |
open source, of distros having their own software repositories and of |
| 97 |
distro maintainers feeding fixes upstream is an exceedingly effective |
| 98 |
combination. Linux systems don't accrue systemic cruft nearly as |
| 99 |
rapidly as Windows systems, in large part because of the forced |
| 100 |
cooperation applied by the LSB and by distro maintainers. |
| 101 |
|
| 102 |
Cruft buildup can still happen, though, and that's why "emerge -e |
| 103 |
@world" exists. And, actually, that's a pretty analogous action to |
| 104 |
reinstalling Windows. It's just much easier, and does a better job of |
| 105 |
retaining user and application settings. |
| 106 |
|
| 107 |
-- |
| 108 |
:wq |
Archives