The GCC Steering Committee changed the next version of GCC from 3.5 to 4.0 while I was in the midst of writing the article. I missed changing a reference; the typo is now fixed.
It fails to point out that compile times are a fairly critical factor in programmer productivity. Yes, we should care that compile times with the Intel compilers look to be over 25% shorter than with gcc on average. (very rough eyeball estimate)
I provide compile times for those benchmarks that ran long enough for a valid measurement to be taken. Clearly GCC is getting slower, and as a Gentoo user, I find that troubling. The GCC development focus is on features and code generation, though there are those who are working diligently on increasing GCC's speed. We'll have to see how GCC 4.0 evolves before its release.
I am amused, also, at how perplexed the author seems that IBM would ship Eclipse 2.x (and CDT 1.x, its associated C/C++ plugins) with the compiler. The CDT plug-ins for Eclipse 3.x were only ready a couple of months ago, so this is plainly a matter of how much time and money the Intel team had to test them with ICC.
As I state in the review, I don't use Eclipse for C or C++ development, nor have I tried the version shipped with Intel's compiler. While Eclipse is reasonable for Java development, it's just too damned slow for me to use in a C environment.
Or maybe they weren't available yet at all: when WAS the Intel 8.1 compiler released, anyway?
GCC 4.0 (which is, of course, "in development") has recently included automatic vectorization -- another of the "good things" that will be coming with the new architecture.
My next article will compare automatic vectorization and profiled optimizations.
When I first wrote reviews for the early geek magazines (anybody remember Microcornucopia?), I could cover a dozen or more compilers for DOS. Today, the field is very thin.:(
The hostility mystifies me. I'm a semi-active participant in the GCC mailing list, and the people who work on GCC are very helpful, open, and communicative. Some even thank me for my QA efforts.
I don't see that my article is negative about GCC; in fact, I'm very clear that Intel's compiler isn't a replacement for GCC, and that GCC is a fine product. Maybe the complainers can't read?
I don't consider your message to be flamebait. There aren't many C and C++ compilers for Linux. Several companies make Fortran 95 compilers for Linux, but few seem interested in the C community.
Fortran 95 offers a number of facilities unavailable in C, C++, or most other languages. This includes some built-in support for parallelism, simplified and powerful array processing, and a lack of aliasing problems. Fortran 95 is structured and modular, and can take advantage of a vast Fortran library comprised of tested, certified, and trusted numerical code.
It's a matter of using the right tool for the job.
Re:why do you use gnome?
on
GNOME 2.8 Released
·
· Score: 2, Interesting
I use both KDE and Gnome (and XFCE from time to time on older boxes). As a developer, I need to see how my programs will operate on both environments, hence my schizophrenic selection of desktop environments.
Gnome applications work and look just fine under KDE; I use Gnumeric as my spreadsheet, but Kword as my word processor. Since the programs run the same way under both desktops, my preference is largely determined by the set-up of the graphical environment. Gnome feels simpler and less "flashy" -- but on higher-end hardware, I like having my KDE eye candy. After all, why have a powerful machine if you don't use it's capabilities?
I recommend Gnome to general users (my wife), while I run KDE most of the time. Like many matters in life, the choice of desktop environment is very personal, and should be base don what works best for you, and not on emotional rants by zealots.
I do a lot of writing -- manuscripts for publication, business documents, software documentation -- so I use many different tools for getting words onto paper.
LaTeX is very utilitarian, and the document sources (being pure text) are eminently portable. But for letters, short documents, and many tasks, I prefer a simple, clean WYSIWYG word processor.
Long ago, on a planet far, far away, I took a liking to the original versions of Microsoft Word -- even the non-graphical version that ran from the MS-DOS command line. It seemed cleaner and more logical than Word Perfect.
Up until a couple of years ago, I used Word under Windows -- but as time passed, I enjoyed using it less and less. Microsoft kept piling on feature after feature; the constant upgrade cycle was frustrating in the extreme. Until just recently, though, "free" and "open" software really didn't provide a good and reliable tool. Today, I have several "free" choices -- and that makes me quite happy.
I'm not fond of OpenOffice. OpenOffice is much too slow on start-up, and it feels almost exactly like Word, but "klunkier". And OpenOffice does not, as of this moment, compile for 64-bit AMD64 (yes, I know I can use 32-bit binaries, but I don't want to).
I like Abiword, though it has bitten me several times with crash bugs. I tend to use Abiword for MS Word documents.
For manuscripts, letters, and most word processing, I've settled on KWord. It starts quick, runs reliably (your mileage may vary), isn't overtly complex, and I have yet to try doing anything that KWord couldn't handle.
On the other hand, for spreadsheets, I've found Gnumeric to be more comfortable than KSpread or OpenOffice Calc.
For me, the appeal of "free" software is choice. I don't really care if other people prefer different solutions -- what I care about is that people can do their work comfortably and reliably. I think companies like Microsoft have forgotten this; they're so wrapped up in trying to force people into upgrades and service contracts, they've lost a sense of building products for people. While "free" software certainly has its problems, I at least get the sense that I'm working with software written by people, not marketroids.
The original Age of Empires includes "priest" units that can "convert" enemy units and buildings -- i.e., causing the target (say, a temple) to change sides. Thus I could take a Persian or Egyptian priest and convert Roman and Greek units to my side.
No matter what you do in life, it will offend someone, somewhere.;)
Small other note: your kids will want to do as you do, so have a plan for introducing them to computers. If Daddy does it all day, they'll want to also.
Yup. I homeschool my three daughters, and they've taken to computers like ducks to water.
I was married seven years, an in my mid-20s when my first daughter. Fifteen years later, I now have three daughters. And I've had mixed experience with jobs and kids.
The most important factor is: Do you work for people who have kids? If not, there will likely to be problems. People who do not have children do not understand the complexities involved; if the school calls with an emergency, a parent has no choice but to respond, even if they're in the middle of a meeting.
Finding a family-friendly employer is difficult; I know this from both personal and friends' experiences. It isn't just a matter of split loyalties -- although that certainly is a factor. Families require insurance and other benedfits; people with kids tend to catch more minor illnesses. Given a choice between a family man (or woman) and someone equally qualified and unattached, the latter often wins. Long ago, families were considered a sign of responsibility and maturity -- today, family is often seen as a burden.
Families are not the only subject of workplace rpejudice. Beyond the obvious "color" and gender issues, religion, hobbies, and even the kind of car you drive can be cause problems in getting hired or staying employed.
I've handled these problems by going solo; this also allows me to homeschool my daughters, and lets me wear shorts in the gawd-awful Florida heat. Not everyone has that luxury, and I'm grateful that my situation allows freedom.
I've written free, open, and closed software; I've accepted money from companies big and small, and I've also been paid to work on non-commercial free projects. I try, very hard, to be an agnostic when it comes to conflicts of religion -- such as the debate of Windows vs. Linux, or KDE vs. Gnome, of C# vs. Java. Tools are tools, and it is how they are used that matters.
And given that case, I do not like how Gnome is being used. Money drives everything, so Novell's involvement only bothers me when I consider their past history of screwing up good things like WordPerfect.
Foremost among my concerns is how Mono can be used as a wedge by Microsoft; if they can not "beat" free software with legal trickery (e.g., SCO), they can slip behind the lines via involvement in Linux via proxies. Microsoft has proven itself untrustworthy, in my personal dealings with them as well as their highly-publicized run-ins with anti-trust law.
I'm not against.NET or C#; I've written some C# code, just as I've written Java and C++ and Python and Fortran and COBOL over the years. I dislike proprietary standards (like.NET and Java) because they trap people, limiting choices and input form the community at large. Give me the chaos of C and C++ over the corporate machinations of C# and Java; I will accept a bit of chaos if my tools can evolve based on conceptual and community diversity.
I normally ignore ACs, but in your case, I'll make an exception.
There have been several debates on this issue on the GCC mailing list; I, and many other numerical users, require -ffast-math for our work. The GCC documentation over-states the issues, in our opinion.
Without the -ffast-math switch, GCC won't emit processor instructions for certain floating point calculations. Without -ffast-math, trig and log functions will be emulated in software. What's the point of having fancy floating-point in your processor if you don't use it?
Now, you can have differences between results on systems that implement different floating-point hardware; if you need deterministic results to the last bit, you probably want to avoid hardware floating-point and go with software routines.
As for accuracy: Testing with industry-standard benchmarks -- such as William Kahan's PARANOIA -- show that using -ffast-math produces more accurate results on some platforms.
Numerical programs do not produce "wrong outputs" when using -ffast-math, nor does that option cause "crashes." You are spreading FUD.
I'd complain about the article's journalistic standards if it had any.
Like so many "reviews"; the topic article is simply one man's diatribe about how a specific product doesn't meet his individual needs. The writer spends more time complaining about Gnome 2.6 than anything else -- and many of those complaints are "taste" issues, based on personal preference, not technical merits.
The beauty of free software is choice. Why trash a product because of personal prejudices, when you can simply try something else? Or, in the case of the reviewer, actually learn how things work, so you can change things to your liking. I run Gnome on one of my workstations; I strongly dislike the spatial browser -- but rather than complain, I changed a setting, and now it works the way I want! Sure, KDE can be "eye candy" heavy at times -- but I turn features off, or, better yet, install a lighter GUI like XFCE on system that don't need the bells and whistle. All this review tells me is that the reviewer is incapable of learning, growing, or changing.
I don't use Fedora, though I have used Red Hat in the past. The choice of distro is very personal and application-specific. I do software development and scientific research; I want bleeding edge and fast performance, so I'm running Gentoo. My wife's laptop, on the other hand, has Mandrake and Windows on it. My servers run Debian.
I've recently made the switch from Debian 'sid' to Gentoo, after frustration with certain Debian policies. I'd previously built a dual Opteron workstation with Gentoo, and found it worked so well that I rebuilt my Pentium 4 workstation with Gentoo as well.
It took 24 hours to completely recompile everything -- base package, KDE, office suites, development tools, Samba -- on a 2.8GHz Pentium 4. I didn;t find this terribly onerous, and the end result is a very clean, fast system. In spite of what some people say, I do see a significant difference in having my code compiled to the hardware it runs on. Heck, I was able to use -ffast-math for the major numerical packages -- try doing *that* with a precompiled distro.:)
I was a Debian user for several years; I still have a dual Pentium 3 and a Sun Ultra 10 running Debian 'sid'. I've used (even paid for) Red Hat, SuSE, Mandrake, and Slackware over the years. And Beyond the time spent compiling, gentoo has been the most pleasant experience yet.
The nice thing about Linxu is that there are so many distros, giving everyone what they want. For me, at this time, Gentoo works very, very well.
An anonymous coward posting a message containing crude insults and arrogant presumption -- this actually makes Mr. Langa look good, in that he is both polite and calm in his reponse to such vitriol.
People buying business or home PCs are not hardware experts -- not should they be expected to be. They want a computer, not a lifestyle. They won't have a clue about chip sets and drivers; expecting them to know such things is like expecting car drivers to know how to machine their own engine parts.
In my not-so-humble opinion, it's the Linux community that looks bad, not Fred Langa. The virulent, dogmatic reponses look childish, especially when they sound like the folks who preach the virtues of tin-foil hats. There are real conspiracies in the universe -- being unable to get sound working with Linux is not one of them.;)
I've been running Linux for a long time, and it's certainly come a long way; seven of nine processors (trekkie pun not intended!) in my office run Linux full-time. And it can be a pain-in-the-rump to install; I've had at least one major hardware problem with every install. Now, once Linux is installed, it offers me many facilities unavailable under Windows -- but then, I'm a developer and engineer, and what I need is quite different than what an office worker or home computist wants.
It's too bad that certain religious fanatics insist upon screaming at heretics and unbelievers when their energies could do so much more for making Linux better.
As someone who has, at times, defending Microsoft, and as someone who has teken Microsoft dollars in the past, I find your leap of assumptions rather disturbing.
I never said Microsoft was evil. Nor is "evil" associated with "public relations" in any dictionary I know of. So what is the basis for your accusations?
Microsoft made a smart, no-risk choice in releasing WTL. How is saying that an insult to Microsoft?
I never said that Microsoft was doing anything evil.
What I am pointing out is that Microsoft took a safe path by releasing something that isn't important to them. It's a good strategic move with few, if any "downsides."
I've written a few WTL programs, and am rather fond of this lightweight wrapper over ATL. WTL allowed me to create small, tight programs without the overhead (and poor design) of MFC (which was created before templates were added to Visual C++).
WTL has always been something of a dirty little secret: it wasn't installed by default by Visual Studio, and the README files suggested that it was an internal project unsupported by management. Yet those of us who disliked MFC found WTL to be quite useful, despite Microsoft's "official" attitude.
As for Microsoft releasing this as Open Source -- MS management never valued WTL in the first place, so "giving it away" is merely a public relations move.
Slashdot Mirror
The GCC Steering Committee changed the next version of GCC from 3.5 to 4.0 while I was in the midst of writing the article. I missed changing a reference; the typo is now fixed.
It's called a "typo". Fixed.
I provide compile times for those benchmarks that ran long enough for a valid measurement to be taken. Clearly GCC is getting slower, and as a Gentoo user, I find that troubling. The GCC development focus is on features and code generation, though there are those who are working diligently on increasing GCC's speed. We'll have to see how GCC 4.0 evolves before its release.
As I state in the review, I don't use Eclipse for C or C++ development, nor have I tried the version shipped with Intel's compiler. While Eclipse is reasonable for Java development, it's just too damned slow for me to use in a C environment.
August 2004.
Ah! A typo! I cut-and-pasted the OS test, and forgot to change "AMD64" to "x86" for the Pentium 4. I just made the trivial edit.
GCC 4.0 (which is, of course, "in development") has recently included automatic vectorization -- another of the "good things" that will be coming with the new architecture.
My next article will compare automatic vectorization and profiled optimizations.
Yes, it is frosting. Chocolate frosting. How else can I maintain my svelt figure?
When I first wrote reviews for the early geek magazines (anybody remember Microcornucopia?), I could cover a dozen or more compilers for DOS. Today, the field is very thin. :(
The hostility mystifies me. I'm a semi-active participant in the GCC mailing list, and the people who work on GCC are very helpful, open, and communicative. Some even thank me for my QA efforts.
I don't see that my article is negative about GCC; in fact, I'm very clear that Intel's compiler isn't a replacement for GCC, and that GCC is a fine product. Maybe the complainers can't read?
I don't consider your message to be flamebait. There aren't many C and C++ compilers for Linux. Several companies make Fortran 95 compilers for Linux, but few seem interested in the C community.
Fortran 95 offers a number of facilities unavailable in C, C++, or most other languages. This includes some built-in support for parallelism, simplified and powerful array processing, and a lack of aliasing problems. Fortran 95 is structured and modular, and can take advantage of a vast Fortran library comprised of tested, certified, and trusted numerical code.
It's a matter of using the right tool for the job.
I use both KDE and Gnome (and XFCE from time to time on older boxes). As a developer, I need to see how my programs will operate on both environments, hence my schizophrenic selection of desktop environments.
Gnome applications work and look just fine under KDE; I use Gnumeric as my spreadsheet, but Kword as my word processor. Since the programs run the same way under both desktops, my preference is largely determined by the set-up of the graphical environment. Gnome feels simpler and less "flashy" -- but on higher-end hardware, I like having my KDE eye candy. After all, why have a powerful machine if you don't use it's capabilities?
I recommend Gnome to general users (my wife), while I run KDE most of the time. Like many matters in life, the choice of desktop environment is very personal, and should be base don what works best for you, and not on emotional rants by zealots.
I do a lot of writing -- manuscripts for publication, business documents, software documentation -- so I use many different tools for getting words onto paper.
LaTeX is very utilitarian, and the document sources (being pure text) are eminently portable. But for letters, short documents, and many tasks, I prefer a simple, clean WYSIWYG word processor.
Long ago, on a planet far, far away, I took a liking to the original versions of Microsoft Word -- even the non-graphical version that ran from the MS-DOS command line. It seemed cleaner and more logical than Word Perfect.
Up until a couple of years ago, I used Word under Windows -- but as time passed, I enjoyed using it less and less. Microsoft kept piling on feature after feature; the constant upgrade cycle was frustrating in the extreme. Until just recently, though, "free" and "open" software really didn't provide a good and reliable tool. Today, I have several "free" choices -- and that makes me quite happy.
I'm not fond of OpenOffice. OpenOffice is much too slow on start-up, and it feels almost exactly like Word, but "klunkier". And OpenOffice does not, as of this moment, compile for 64-bit AMD64 (yes, I know I can use 32-bit binaries, but I don't want to).
I like Abiword, though it has bitten me several times with crash bugs. I tend to use Abiword for MS Word documents.
For manuscripts, letters, and most word processing, I've settled on KWord. It starts quick, runs reliably (your mileage may vary), isn't overtly complex, and I have yet to try doing anything that KWord couldn't handle.
On the other hand, for spreadsheets, I've found Gnumeric to be more comfortable than KSpread or OpenOffice Calc.
For me, the appeal of "free" software is choice. I don't really care if other people prefer different solutions -- what I care about is that people can do their work comfortably and reliably. I think companies like Microsoft have forgotten this; they're so wrapped up in trying to force people into upgrades and service contracts, they've lost a sense of building products for people. While "free" software certainly has its problems, I at least get the sense that I'm working with software written by people, not marketroids.
The original Age of Empires includes "priest" units that can "convert" enemy units and buildings -- i.e., causing the target (say, a temple) to change sides. Thus I could take a Persian or Egyptian priest and convert Roman and Greek units to my side.
No matter what you do in life, it will offend someone, somewhere. ;)
Yup. I homeschool my three daughters, and they've taken to computers like ducks to water.
Okay, I know that first sentence was correct when I hit "Submit!" It should read:
I was married seven years, and in my mid-20s when my first daughter was born.
Grumble, grumble...
I was married seven years, an in my mid-20s when my first daughter. Fifteen years later, I now have three daughters. And I've had mixed experience with jobs and kids.
The most important factor is: Do you work for people who have kids? If not, there will likely to be problems. People who do not have children do not understand the complexities involved; if the school calls with an emergency, a parent has no choice but to respond, even if they're in the middle of a meeting.
Finding a family-friendly employer is difficult; I know this from both personal and friends' experiences. It isn't just a matter of split loyalties -- although that certainly is a factor. Families require insurance and other benedfits; people with kids tend to catch more minor illnesses. Given a choice between a family man (or woman) and someone equally qualified and unattached, the latter often wins. Long ago, families were considered a sign of responsibility and maturity -- today, family is often seen as a burden.
Families are not the only subject of workplace rpejudice. Beyond the obvious "color" and gender issues, religion, hobbies, and even the kind of car you drive can be cause problems in getting hired or staying employed.
I've handled these problems by going solo; this also allows me to homeschool my daughters, and lets me wear shorts in the gawd-awful Florida heat. Not everyone has that luxury, and I'm grateful that my situation allows freedom.
I've written free, open, and closed software; I've accepted money from companies big and small, and I've also been paid to work on non-commercial free projects. I try, very hard, to be an agnostic when it comes to conflicts of religion -- such as the debate of Windows vs. Linux, or KDE vs. Gnome, of C# vs. Java. Tools are tools, and it is how they are used that matters.
And given that case, I do not like how Gnome is being used. Money drives everything, so Novell's involvement only bothers me when I consider their past history of screwing up good things like WordPerfect.
Foremost among my concerns is how Mono can be used as a wedge by Microsoft; if they can not "beat" free software with legal trickery (e.g., SCO), they can slip behind the lines via involvement in Linux via proxies. Microsoft has proven itself untrustworthy, in my personal dealings with them as well as their highly-publicized run-ins with anti-trust law.
I'm not against .NET or C#; I've written some C# code, just as I've written Java and C++ and Python and Fortran and COBOL over the years. I dislike proprietary standards (like .NET and Java) because they trap people, limiting choices and input form the community at large. Give me the chaos of C and C++ over the corporate machinations of C# and Java; I will accept a bit of chaos if my tools can evolve based on conceptual and community diversity.
I normally ignore ACs, but in your case, I'll make an exception.
There have been several debates on this issue on the GCC mailing list; I, and many other numerical users, require -ffast-math for our work. The GCC documentation over-states the issues, in our opinion.
Without the -ffast-math switch, GCC won't emit processor instructions for certain floating point calculations. Without -ffast-math, trig and log functions will be emulated in software. What's the point of having fancy floating-point in your processor if you don't use it?Now, you can have differences between results on systems that implement different floating-point hardware; if you need deterministic results to the last bit, you probably want to avoid hardware floating-point and go with software routines.
As for accuracy: Testing with industry-standard benchmarks -- such as William Kahan's PARANOIA -- show that using -ffast-math produces more accurate results on some platforms.
Numerical programs do not produce "wrong outputs" when using -ffast-math, nor does that option cause "crashes." You are spreading FUD.
I'd complain about the article's journalistic standards if it had any.
Like so many "reviews"; the topic article is simply one man's diatribe about how a specific product doesn't meet his individual needs. The writer spends more time complaining about Gnome 2.6 than anything else -- and many of those complaints are "taste" issues, based on personal preference, not technical merits.
The beauty of free software is choice . Why trash a product because of personal prejudices, when you can simply try something else? Or, in the case of the reviewer, actually learn how things work, so you can change things to your liking. I run Gnome on one of my workstations; I strongly dislike the spatial browser -- but rather than complain, I changed a setting, and now it works the way I want! Sure, KDE can be "eye candy" heavy at times -- but I turn features off, or, better yet, install a lighter GUI like XFCE on system that don't need the bells and whistle. All this review tells me is that the reviewer is incapable of learning, growing, or changing.
I don't use Fedora, though I have used Red Hat in the past. The choice of distro is very personal and application-specific. I do software development and scientific research; I want bleeding edge and fast performance, so I'm running Gentoo. My wife's laptop, on the other hand, has Mandrake and Windows on it. My servers run Debian.
Freedom is about choice.
I've recently made the switch from Debian 'sid' to Gentoo, after frustration with certain Debian policies. I'd previously built a dual Opteron workstation with Gentoo, and found it worked so well that I rebuilt my Pentium 4 workstation with Gentoo as well.
It took 24 hours to completely recompile everything -- base package, KDE, office suites, development tools, Samba -- on a 2.8GHz Pentium 4. I didn;t find this terribly onerous, and the end result is a very clean, fast system. In spite of what some people say, I do see a significant difference in having my code compiled to the hardware it runs on. Heck, I was able to use -ffast-math for the major numerical packages -- try doing *that* with a precompiled distro. :)
I was a Debian user for several years; I still have a dual Pentium 3 and a Sun Ultra 10 running Debian 'sid'. I've used (even paid for) Red Hat, SuSE, Mandrake, and Slackware over the years. And Beyond the time spent compiling, gentoo has been the most pleasant experience yet.
The nice thing about Linxu is that there are so many distros, giving everyone what they want. For me, at this time, Gentoo works very, very well.
An anonymous coward posting a message containing crude insults and arrogant presumption -- this actually makes Mr. Langa look good, in that he is both polite and calm in his reponse to such vitriol.
People buying business or home PCs are not hardware experts -- not should they be expected to be. They want a computer, not a lifestyle. They won't have a clue about chip sets and drivers; expecting them to know such things is like expecting car drivers to know how to machine their own engine parts.
In my not-so-humble opinion, it's the Linux community that looks bad, not Fred Langa. The virulent, dogmatic reponses look childish, especially when they sound like the folks who preach the virtues of tin-foil hats. There are real conspiracies in the universe -- being unable to get sound working with Linux is not one of them. ;)
I've been running Linux for a long time, and it's certainly come a long way; seven of nine processors (trekkie pun not intended!) in my office run Linux full-time. And it can be a pain-in-the-rump to install; I've had at least one major hardware problem with every install. Now, once Linux is installed, it offers me many facilities unavailable under Windows -- but then, I'm a developer and engineer, and what I need is quite different than what an office worker or home computist wants.
It's too bad that certain religious fanatics insist upon screaming at heretics and unbelievers when their energies could do so much more for making Linux better.
As someone who has, at times, defending Microsoft, and as someone who has teken Microsoft dollars in the past, I find your leap of assumptions rather disturbing.
I never said Microsoft was evil. Nor is "evil" associated with "public relations" in any dictionary I know of. So what is the basis for your accusations?
Microsoft made a smart, no-risk choice in releasing WTL. How is saying that an insult to Microsoft?
I never said that Microsoft was doing anything evil.
What I am pointing out is that Microsoft took a safe path by releasing something that isn't important to them. It's a good strategic move with few, if any "downsides."
I've written a few WTL programs, and am rather fond of this lightweight wrapper over ATL. WTL allowed me to create small, tight programs without the overhead (and poor design) of MFC (which was created before templates were added to Visual C++).
WTL has always been something of a dirty little secret: it wasn't installed by default by Visual Studio, and the README files suggested that it was an internal project unsupported by management. Yet those of us who disliked MFC found WTL to be quite useful, despite Microsoft's "official" attitude.
As for Microsoft releasing this as Open Source -- MS management never valued WTL in the first place, so "giving it away" is merely a public relations move.