Jump to content

Recommended Posts


Posted

What have Microsoft's tens of thousands of engineers been doing?

 

First thing we wish to thank you for this question. :)

 

Besides changing light bulbs ;) and tagging as by design anything they had no idea how to fix :

http://reboot.pro/topic/3541-how-many-microsoft-programmers-does-it-take-to-screw-in-a-light-bulb/

and publishing either misleading or plainly wrong documentation:

http://www.msfn.org/board/topic/170497-whats-it-going-to-take-to-restore-previous-versions-from-shadow-copies/page-2#entry1086058

(and shuffling it a lot around their site), a substantial amount of them used last 13 (thirteen) years to solve (maybe) a small issue created when - in the course of bettering the Operating System - they managed to worsen it by removing support for an already existing feature:

http://www.cl.cam.ac.uk/~mgk25/mswish/ut-rtc.html

 

jaclaz

Posted

 

Why aren't our computers capable of speech recognition and very good synthesis?  Why don't we have spam protection that works any better than in 2003?  Why can't the system remember where every application window was last time we used it?  Why can't we describe an image or music in terms we'd normally use and have the system search for it (and actually find it)?

 

The best thing about it is that ten years ago we were closer to this than now - Remember the rumours about the wonders of WinFS?

Posted

EXACTLY.  It's as though someone suddenly turned off a switch labeled "care about serious computing".

 

Was it coincident with Bill Gates turning his attention away and toward eliminating malaria?  Say what you will, he IS a geek, and geeks have more noble motives than most jocks. 

 

Does this say it's improbable a high tech company can maintain it's original charter and course without the original visionary at the helm?

 

-Noel

Posted

EXACTLY.  It's as though someone suddenly turned off a switch labeled "care about serious computing".

 

Well, but you have to recognize that one can at the same time provide better anti-bot protection (serious computing) and help abandoned animals (be involved in bettering the society) :w00t: .

http://research.microsoft.com/en-us/um/redmond/projects/asirra/

 

And this also answers partially the previous questions about what some of the good MS guys have been up to lately ... ;)

 

Though of course cats (particularly lolcats) are much more popular on the internet than dogs (or loldogs), the question is clearly biased towards cats as it is fixed into:

”Please select all the cat photos:”

 

Additionally (to show how international is the approach, and how the culture of attention to details is rooted in Redmond ) it is "fixed" in English:

http://www.mediawiki.org/wiki/Extension_talk:Asirra 

 

And now, for NO apparent reason:

 

Pick-Your-Battles-Best-Demotivational-Po

 

jaclaz

Posted

Though of course cats (particularly lolcats) are much more popular on the internet than dogs (or loldogs), the question is clearly biased towards cats as it is fixed into:

”Please select all the cat photos:”

Hmmmm well I know of no lolcat coins or lolcat race cars...

Yet we have those for loldogs now:

http://www.nascar.com/en_us/news-media/blogs/Off-Track/doge-reddit-josh-wise-talladega-superspeedway-aarons-499.html

;)

Posted

 

What have Microsoft's tens of thousands of engineers been doing?

 

First thing we wish to thank you for this question. :)

 

Besides changing light bulbs ;) and tagging as by design anything they had no idea how to fix :

http://reboot.pro/topic/3541-how-many-microsoft-programmers-does-it-take-to-screw-in-a-light-bulb/

and publishing either misleading or plainly wrong documentation:

http://www.msfn.org/board/topic/170497-whats-it-going-to-take-to-restore-previous-versions-from-shadow-copies/page-2#entry1086058

(and shuffling it a lot around their site), a substantial amount of them used last 13 (thirteen) years to solve (maybe) a small issue created when - in the course of bettering the Operating System - they managed to worsen it by removing support for an already existing feature:

http://www.cl.cam.ac.uk/~mgk25/mswish/ut-rtc.html

 

jaclaz

 

 

The stuff on that last link comes off to me as neurotic hand-wringing:

 

Universal Time can be unambiguously converted into a local time. The converse is not true, that is a local time cannot reliably be converted back into Universal time. This is due to the Summer Time or Daylight Savings Time offset periods implemented in many countries. At the end of that offset period, local-time clocks have to be turned back by usually one hour, therefore a 60 minute period on the local-time scale is repeated. There are two possible corresponding hours on the Universal Time scale for the repeated local-time perios, which makes it impossible to unambiguously determine Universal Time during that hour. This obviously makes local time a bad choice for the time format maintained in an independently-running hardware time base. If a computer is booted during the hour after DST ends, it cannot determine UT reliably.

 

I don't know about other countries, but in the U.S. the time change takes place at 2AM local ;) time. How many computers ever need to be rebooted at 2 in the morning??

 

 

Users will not have to be bothered any more with pop-up queries that ask whether the RTC clock should be updated. With a Universal Time RTC, Windows machines can now reliably update their local time at exactly the right time without any delay and the need for manual user intervention, which is especially critical for reliable unsupervised operation in embedded systems, data centers, etc.

 

I haven't had to manually update a computer clock for daylight savings since I moved my work from Windows 98 to Vista.

 

 

If the hardware clock runs in UTC, there is no need any more to keep timezone-related functionality in the Windows kernel, as the need to perform a local-time to UTC conversion during boot or after hibernation falls away. Simpler is better!

 

But if the hardware clock runs in UTC, then for the sake of the user the OS will have to convert UTC into local time. How is that simpler? A conversion still needs to take place. In any case, if I'm in New Delhi, why would I need my PC to convert from UTC to New Delhi time at any level (BIOS or OS)? I'd rather have everything shown in my local time. UTC doesn't come into play at all and there's no need to convert anything in either direction.

 

People in certain specialized computing environments might have a use (although I can't think of any) for what the guy proposes, but for 99.999% of users out there, it seems to me that running the BIOS clock in UTC is not only unnecessary, but also a source of potential confusion (and maybe even alarm) at those times when they do need to go into the BIOS for whatever reason, and then see a different time than they expect.

 

Maybe I just need my morning coffee :boring: , but to me this really seems to be a solution in search of a problem.

 

--JorgeA

Posted

Maybe I just need my morning coffee :boring: , but to me this really seems to be a solution in search of a problem.

 

You'd better get a double one today, strong and black, as seemingly you need it. :yes:

 

Time/Date (mis)management has been a traditional issue since the advent of computers, and the number of issues (minor or major) that have been caused by it worldwide are slightly more relevant than what you might have experienced in your local (US) and personal (home) usage of a computer.

 

 

Consider how FAT (and other filesystems, like CDFS ) save date/time in "local time" (CDFS has additionally a field for offset from Greenwich) whilst NTFS saves them in UTC, and how in the past there were several cases of issues with Windows based computers keeping date/time, a few ones:

http://support.microsoft.com/kb/909915/en-us

http://support.microsoft.com/kb/2400550/en-us

 

March 2007 has been a busy month on the DEDICATED BLOG that MS had to put online :whistle::

http://blogs.technet.com/b/dst2007/archive/2007/03.aspx

for an astounding number of issues which not necessarily are "Microsoft's fault" as a number of them are connected to changes in local Laws, the point made in the article is that for a number of reasons the "conversion" may fail and that the "design" of the whole thing is broken, as you have an automated manipulation of the RTC that may easily fail:

http://blogs.technet.com/b/dst2007/archive/2007/03/11/time-zones-not-updating-on-windows-2003-windows-xp-or-windows-2000.aspx

 

More or less *anything* that has a need for "synchronization" or that however relies on date/time to perform some automated action is subject (thanks to Murphy's Law) to an increased risk of failing.

 

A case where a dual boot system caused a mis-setting of the RTC clock:

http://catless.ncl.ac.uk/Risks/22.34.html#subj3.1

 

And right following a case about the known possible issues with cross-platform filestamps local or through a remote connection: 

http://support.microsoft.com/default.aspx?scid=KB;EN-US;Q129574

 

Surely these issues are not very common for the "average Joe" and most of them are avoided by "solid" programming/scheduling/procedures by the IT managers of international firms, but still it is (was) a completely wrong approach, particularly when it is "common" between "home" OS's and Server OS's.

 

jaclaz

Posted

Thanks for the details, jaclaz.

 

I see that three of the issues affect Microsoft OSes as late as Windows Server 2003 and Windows XP. Another problem discussion dates from 2002 and refers to one of those three issues. Another link shows issues that were live "over 7 years ago."

 

That leaves one which involved Vista and Windows 7, and that one "can occur" if

 

the latest Windows Daylight Saving Time Cumulative Update has not been applied.

 

I dunno, it just doesn't seem like a big deal, at least not any more.

 

--JorgeA

 

Posted

 

I dunno, it just doesn't seem like a big deal, at least not any more.

 

Not that I presented it as such:

 

(and shuffling it a lot around their site), a substantial amount of them used last 13 (thirteen) years to solve (maybe) a small issue created when - in the course of bettering the Operating System - they managed to worsen it by removing support for an already existing feature:

http://www.cl.cam.ac.uk/~mgk25/mswish/ut-rtc.html

 

The point here is that the "original" (wrong, but easily fixable) approach of "forcing" the user to keep the RTC on "local time" took all of more than 7 years to be addressed (BTW, the provision had ALREADY been there in Windows NT but was "forgotten"), then once the (simple, small) bug was fixed, it was re-introduced twice or thrice.

 

The point being that the actual cause of the problem emerged in the early days of computing only when the PC started to get "interconnected" and was addressed by the POSIX standard in a more than satisfying manner.

 

For whatever reasons the good MS guys did not conform (or did not conform fully) to it and continued using the (flawed) MS-DOS approach (which would have been alright in itself had everything being coded "right" and there weren't the - already known - needs for interconnection).

 

When finally the possibility of having the RTC set to GMT was given (with Vista SP2 and Windows 7) thus solving a whole set of small possible issues, MS repeatedly created in updates further issues (possibly even worse than the original ones).

 

Since date and time are in themselves a precious part of the computing world one would have expected from MS some more attention and care  to the specific implementations.

 

Just for the fun of it, I will provide you a nice link:

http://stackoverflow.com/questions/4967903/linux-windows-timezone-mapping

check particularly the "Obligatory Time Zone Rant:"

which may lead you to here:

http://en.wikipedia.org/wiki/Wikipedia:Reference_desk/Archives/Miscellaneous/2008_April_22#Romance_Standard_Time

which hints how the whole "Romance Standard Time" might be a quiet invention of the good MS guys :whistle:

 

More seriously, the good MS guys are not alone in making a mess of this specific topic, as there are (scarcely populated) timezones that do not actually exist :w00t:

http://www.forensicfocus.com/Forums/viewtopic/t=11879/

 

jaclaz

Posted (edited)

That IS a nice link, thanks. It led me to a bunch of other oddities.  :)

 

You're right, time-zone issues aren't limited to Microsoft.

 

Maybe the simplest solution is to put the whole world under UTC, so that wherever you are located it'll be the same clock hour... even if it means that 14:00 will be in plain daylight in one place and in the middle of the night somewhere else. And presto! no more time-zone difficulties. :angel

 

IMHO clock tricks like changing from standard to daylight savings time need to be abolished, too. It leads to so much confusion, and not just of the computer kind.

 

Although it does have its benefits sometimes: I remember a few years ago, some terrorists died when their car bomb went off an hour early because they forgot to factor in the "spring forward" time change.

 

Other time-zone weirdness includes the "Further-Eastern :crazy: European Time" in Kaliningrad (Koenigsberg) where it can be an hour ahead of countries that lie to the east; or "Iran Standard Time" that's a half-hour different from countries directly to the north and south; and Venezuela, which put itself a half-hour behind the Caribbean islands straight north of it. There are even some time zones that are 15 or 45 minutes off from the neighboring areas. :wacko:

 

--JorgeA

Edited by JorgeA
Posted

IMHO clock tricks like changing from standard to daylight savings time need to be abolished, too. It leads to so much confusion, and not just of the computer kind.

Well, the good ol'  Old Indian (apocryphal) example of the blanket is not that bad, JFYI:

daylight.jpg

 

Consider also the "other" possible consequences in practice ;):

pets+DST.jpeg

 

and, in order to avoid a possible accusation (actually true) of being biased:

4055754_orig.jpg

 

jaclaz

Posted (edited)

This time zone talk does illustrate another rather more serious issue, though...

 

As we progress through time we're expected to advance the state of the art.  Everything engineered today is supposed to be better in every way than everything engineered, yesterday, last month, last century...

 

No one would think a car made exactly as new cars were made in the 1970s is acceptable.  Remember those, that rusted-out in about 3 or 4 years?  That killed their occupants in the most minor of collisions?

 

To advance the state of the art requires that a great deal more thought be put into things now before deciding to build.  To some extent, things are codified (e.g., building codes) to make sure we don't repeat the mistakes of the past, but not really for commercial stuff. 

 

It's expected that things get better and better.

 

And in general they have.

 

But there have been hiccups along the way.  From the stuff of legend to everyday stuff, occasionally we humans demonstrate our natural, almost overwhelming tendency to do stupid work rather than good work.

 

Remember the wonder of cell phones?  Everyone naturally assumed that they would work as well as a real, wired phone (heavily governed by law) and would just be wireless.  In reality...  When have you heard a conversation by someone on a phone where they didn't spend a lot of time asking people to repeat themselves, or outright talking about how bad the connection was?  That technology has been "almost working" for how long now?

 

And now we have operating systems.

 

For a while, things seemed to get better and better.

 

For a while.

 

-Noel

Edited by NoelC
Guest
This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...