Those are the best kinds of bugs. Something repeatable. Nothing 4,000 debug statements added one after another cannot find and fix... Just takes time and a SHITLOAD of patience...
Too bad I'm not a kernel hacker, I'm good at fixing these.:-/
How does one go about doing this? I've got a number of apps on deployed webservers all over the place that are generally the same (libraries and distributions, SuSE, FC2, FC4) and can save myself a shitload of time by building customized RPM's for my feature sets, without having to recompile everywhere.
I have a particular apache/perl/php4/subversion config I like... is it as simple as running rpmbuild on said tarfile?
Two words: lower taxes. If all this "I have to spend it, or next year I'll get crap for budget" would stop, we wouldn't be giving Uncle Sam %25-40 of our pay.
It leads to shit like I-90 in Massachusetts getting resurfaced every 10 fscking years...
Take outlook 2003 configured to Exchange Server 2000. Install Sales Logix 6.2. Tell me how you fix the ensuing registry bug without actually touching the registry. Maybe it's just my environment but all my SalesLogix users are getting registry corruption... not major, but certainly beyond 99% of them.
4 boxes in a cluster, running a mix of services, each one capable of running *ALL* of them if the other three go down (in degraded performance mode, obviously). This is sometimes difficult with Microsoft, nevermind the software cost of Datacenter Server...
More power usage if services are all on separate boxes, but I think the maintenance requirements to are roughly the same. Doesn't take much on any platform to log in and run up2date/windowsupdate.
I certainly cannot agree that MS would never do this. In fact, it would seem that it's an obvious enhancement, one in fact that I think they'll work on for their next general OS release (Windows 2007 home, or whatever).
The problem with writing something for Windows is vendor acceptance. RPM and DEB are pretty standardized in the Unix world, broken though they are. Getting something universal that works in Windows is harder. We should instead be pressuring Microsoft (as developers) to open up Windows Update in the name of all that is good and proper.
Problem is, Windows Update is very tightly integrated with Microsoft.com, and I'm not sure they'll fix that.
My hope is that people wouldn't have to visit Windows Update to get updates, but that Windows update becomes truly integrated with Windows. Right now, it's a simple app that speaks with microsoft.com via HTTP. It sure could be adapted to accept instructions to talk to say update.symantec.com or update.winzip.com or update.whoever.com to get update info.
No references, just a guess based on their past history. Things enter the OS, and eventually get exposed as APIs to the rest of us. I imagine it's just a matter of time before someone (oracle, etc) asks MS, hey, why can't you open WindowsUpdate so that it'll check this registered address for updates? Look at sparse files, a feature added to NTFS specifically for SQL Server circa NT 4 SP3 or 4. Big secret, got everyone all pissed off so that MS committed to no more feature upgrades rolled into service packs.
And it's possible that the answer is that Microsoft won't, because there's a lot of backend implementation that needs to be done on the update site. I surely cannot imagine a simple webservice that takes a string with the windows version, service packs and all, and the specific app registration data the user enters (serial numbers, etc) to run a quick update check. With the patch data, you can even specifically warn the user before the update that it is incompatible with their system, and why.
Microsoft has an opportunity here, it remains to be seen whether they will embrace it.
Patience... soon enough Windows Update will become an API you can leverage as an app developer to tell Windows where to find patches and updates to your software. Just wait.
Linux doesn't have a problem. In fact, in units shipped, I'd gather it's nipping Microsoft's heels for most popular OS of all time (embedded devices included).
It's the DISTRIBUTIONS that need to collaborate. Yast/RPM/Deb/YUM/apt, etc. KDE/Gnome Ick. Life was bad enough when Motif broke off from standard X. My guess in the long haul is gnome (evolution, vmware, other business apps are GTK). I have yet to run a gnome app that gives me a top-of-the-z-order modal splash screen (kdemail, k3b, others).
Yup, it's the distributions holding Linux back. Standardized installer/packager. LSB compliance. Desktop compliance (baseline). Supporting KDE *AND* Gnome sucks.
A true live sorta Watcom employee? OMG. I'm blushing. Watcom C++ 9 was my gateway in to flat memory model heaven, Dos Windows AND OS/2. I've saved that CD. I'm going to turn it into a plaque someday...
Then I got corrupted by the borg (VC++). MSDN Library is just so irresistable.
Why should downloading a torrent of WindowsXP be illegal if I've already paid for the license?
Distributing on the other hand... mkay, got me there... but simple acquisition? I suppose I could pay $25 for the CD in the mail, but what if I need it in a hurry, on a weekend, when Insight or CDW aren't open?
Just curious is all... thought I'd throw out some trollbait.:-)
We know slashdot can survive a slashdotting... Is/. the basis by which all bandwidth is judged? Should we start measuring bandwith in Slashdot units now? Is Libraries of Congress no longer applicable? Is there a conversion factor? 1/. == xxx LoC?
Yea, um, if you don't have drivers to actually boot the disk, you need a driver disk, to install said devices and format.
Case in point, I had to install an HP Blade BL20 G3 (brand new P4) yesterday, and couldn't get the Windows 2000 boot process to see the drives (there are USB CDROM and Floppy drives on this, since blades don't typically have any of those). And it wouldn't work. HP ships a specially configured CD for installing Windows.
So yea, it's possible he's running into a problem. Not supporting driver CD's is a problem Microsoft needs to solve. Most linuxes I use support manufacturer disks that aren't floppies... Why can't WindowsXP?
Or mirrors, without the abject explicit links we have now... If the browser detects a mirror isn't working, it cycles to the next link in the loop, ad infinitum until they all time-out or the content gets delivered.
Slashdot Mirror
A couple years back?? WTF are you smoking. They haven't made Nova's in over twenty years!!!
Those are the best kinds of bugs. Something repeatable. Nothing 4,000 debug statements added one after another cannot find and fix... Just takes time and a SHITLOAD of patience...
:-/
Too bad I'm not a kernel hacker, I'm good at fixing these.
How does one go about doing this? I've got a number of apps on deployed webservers all over the place that are generally the same (libraries and distributions, SuSE, FC2, FC4) and can save myself a shitload of time by building customized RPM's for my feature sets, without having to recompile everywhere.
I have a particular apache/perl/php4/subversion config I like... is it as simple as running rpmbuild on said tarfile?
Or a compiler.
Two words: lower taxes. If all this "I have to spend it, or next year I'll get crap for budget" would stop, we wouldn't be giving Uncle Sam %25-40 of our pay.
It leads to shit like I-90 in Massachusetts getting resurfaced every 10 fscking years...
Take outlook 2003 configured to Exchange Server 2000. Install Sales Logix 6.2. Tell me how you fix the ensuing registry bug without actually touching the registry. Maybe it's just my environment but all my SalesLogix users are getting registry corruption... not major, but certainly beyond 99% of them.
4 boxes in a cluster, running a mix of services, each one capable of running *ALL* of them if the other three go down (in degraded performance mode, obviously). This is sometimes difficult with Microsoft, nevermind the software cost of Datacenter Server...
More power usage if services are all on separate boxes, but I think the maintenance requirements to are roughly the same. Doesn't take much on any platform to log in and run up2date/windowsupdate.
If everyone was using NTP, this wouldn't be a problem.
Good enough to sig... :-)
I certainly cannot agree that MS would never do this. In fact, it would seem that it's an obvious enhancement, one in fact that I think they'll work on for their next general OS release (Windows 2007 home, or whatever).
The problem with writing something for Windows is vendor acceptance. RPM and DEB are pretty standardized in the Unix world, broken though they are. Getting something universal that works in Windows is harder. We should instead be pressuring Microsoft (as developers) to open up Windows Update in the name of all that is good and proper.
Problem is, Windows Update is very tightly integrated with Microsoft.com, and I'm not sure they'll fix that.
My hope is that people wouldn't have to visit Windows Update to get updates, but that Windows update becomes truly integrated with Windows. Right now, it's a simple app that speaks with microsoft.com via HTTP. It sure could be adapted to accept instructions to talk to say update.symantec.com or update.winzip.com or update.whoever.com to get update info.
As if they don't already?
No references, just a guess based on their past history. Things enter the OS, and eventually get exposed as APIs to the rest of us. I imagine it's just a matter of time before someone (oracle, etc) asks MS, hey, why can't you open WindowsUpdate so that it'll check this registered address for updates? Look at sparse files, a feature added to NTFS specifically for SQL Server circa NT 4 SP3 or 4. Big secret, got everyone all pissed off so that MS committed to no more feature upgrades rolled into service packs.
And it's possible that the answer is that Microsoft won't, because there's a lot of backend implementation that needs to be done on the update site. I surely cannot imagine a simple webservice that takes a string with the windows version, service packs and all, and the specific app registration data the user enters (serial numbers, etc) to run a quick update check. With the patch data, you can even specifically warn the user before the update that it is incompatible with their system, and why.
Microsoft has an opportunity here, it remains to be seen whether they will embrace it.
Patience... soon enough Windows Update will become an API you can leverage as an app developer to tell Windows where to find patches and updates to your software. Just wait.
ReiserFS, XFS, JFS2, Ext3. Basically all the currently used ones.
Linux doesn't have a problem. In fact, in units shipped, I'd gather it's nipping Microsoft's heels for most popular OS of all time (embedded devices included).
It's the DISTRIBUTIONS that need to collaborate. Yast/RPM/Deb/YUM/apt, etc. KDE/Gnome Ick. Life was bad enough when Motif broke off from standard X. My guess in the long haul is gnome (evolution, vmware, other business apps are GTK). I have yet to run a gnome app that gives me a top-of-the-z-order modal splash screen (kdemail, k3b, others).
Yup, it's the distributions holding Linux back. Standardized installer/packager. LSB compliance. Desktop compliance (baseline). Supporting KDE *AND* Gnome sucks.
A true live sorta Watcom employee? OMG. I'm blushing. Watcom C++ 9 was my gateway in to flat memory model heaven, Dos Windows AND OS/2. I've saved that CD. I'm going to turn it into a plaque someday...
Then I got corrupted by the borg (VC++). MSDN Library is just so irresistable.
Why should downloading a torrent of WindowsXP be illegal if I've already paid for the license?
:-)
Distributing on the other hand... mkay, got me there... but simple acquisition? I suppose I could pay $25 for the CD in the mail, but what if I need it in a hurry, on a weekend, when Insight or CDW aren't open?
Just curious is all... thought I'd throw out some trollbait.
Right, forgot, the host header is in zee encrypted packet.. I knew it was something stupid like that...
We know slashdot can survive a slashdotting... Is /. the basis by which all bandwidth is judged? Should we start measuring bandwith in Slashdot units now? Is Libraries of Congress no longer applicable? Is there a conversion factor? 1 /. == xxx LoC?
Um, please please don't prevent ANYONE from upgrading from WinME. Please? Every now and again that little bastard shows up and terrorizes me...
:-)
I had lots of great fun this weekend using all the MSDN WindowsME cds I have as skeet targets.
Why would Intel care what OS you point on your machine?
Yea, um, if you don't have drivers to actually boot the disk, you need a driver disk, to install said devices and format.
Case in point, I had to install an HP Blade BL20 G3 (brand new P4) yesterday, and couldn't get the Windows 2000 boot process to see the drives (there are USB CDROM and Floppy drives on this, since blades don't typically have any of those). And it wouldn't work. HP ships a specially configured CD for installing Windows.
So yea, it's possible he's running into a problem. Not supporting driver CD's is a problem Microsoft needs to solve. Most linuxes I use support manufacturer disks that aren't floppies... Why can't WindowsXP?
I can find a HUNDRED USB keychains/storage devices in my office, but I can't find one fscking floppy disk...
Not sure if this would work, but in Apache, would the following work:
NameVirtualHost ipaddr:443
<virtualhost ipaddr:443>
ServerName www.website.com
<ssl config>
</ssl>
</virtualhost>
<virtualhost ipaddr:443>
ServerName www.website2.com
<ssl config>
</ssl>
</virtualhost>
I've only ever done IP based SSL, I've never gone to the trouble of trying to make it work with VHosts.
Or mirrors, without the abject explicit links we have now... If the browser detects a mirror isn't working, it cycles to the next link in the loop, ad infinitum until they all time-out or the content gets delivered.