There is no turn around in Dell. And there was better Linux/BSD support for Linux *before* Dell supported Linux.
We've been a Dell customer for 10 years. The first five years were pretty good. The old PowerEdge servers were pretty nice. But I've never met a Perc card I liked.
What is a Perc card? Well, it is generic term for the card that Dell happens to be rebranding at the moment.
What is rebranding? Well, take a perhaps decent card, screw with it for the sake of vendor lock-in, so that the original vendor's drivers and utilities no longer work with it, degrade its performance beyond measure without any hope of cure, and you have a Dell rebrand.
The new MD1000, while pretty nice arrays, are only officially supported if you use a Perc 5e card. This card is a true Dell rebrand of an LSI MegraRAID. It's a thousand dollar turd. There are no decent utilities (in fact all LSI megaraid utils will NOT work). They even took away some of the simple command line utilities for Linux like lsiutil, and replaced it with a huge, bloated Java application. Can you imagine that piece of bloatware as your only means of talking to your drive array? And FreeBSD support, while the FreeBSD community has made a valiant effort trying to support this piece of shit, still is not worth it. It's lackluster performance at best.
But here's the real kicker about the Perc 5e: It does not allow direct pass through. Yeah, that's right, if you get fed up with it's lame RAID 10 performance (and it is lame we've confirmed that software RAID on Linux and FreeBSD is significantly faster), you can't directly reach your drives. So if you want to implement software RAID, you have to configure independent RAID0 virtual disks in the Perc 5e BIOS. Why? so Dell can lock you in to their proprietary, rebranded, turd card. This is a wonderful new feature unique to the Perc 5 that wasn't there with Perc 4.
But, as we found out, the Perc 4 was an even bigger piece of crap. It was a rebranded Adaptec that had over a year-long history of locking up under heavy loads on Linux systems. They were selling this so-called Linux supported PowerEdge 2650 (boat anchor) for over a year knowing that it locked up under stress:
https://bugzilla.redhat.com/show_bug.cgi?id=92129
We got burned by this and it took us weeks to resolve (went through 3 Perc 4 cards with same result, and ended up putting in a LSI card).
If you need anything with real RAID support, run from Dell. We've been pretty pleased with the MD1000, but that Perc 5 card sums up Dell. Buy a 3ware or Areca instead. Who cares if Dell won't support the MD1000 if you use it, they don't support it anyway. Of all our calls to Dell enterprise support, we've talked to one person who seems to have half of a clue.
There is no such thing as Dell support. And their lame attempt at vendor lock-in only makes it worse. And that is why Dell is drowning.
We've moved to HP. And we've been really happy with our HP servers. Is HP support any better? I can't say. But we don't have near the vendor lock-in and rebranded voodoo we get from Dell.
I have a masters in chemical engineering. I come from a family of engineers: my dad's an engineer, his dad was one, and my father in law is one -- altogether we cover electrical, civil, chemical, and industrial.
Here's my take on why so few Americans are going into engineering:
Engineers are treated like shit.
While the starting salaries are okay, long term growth sucks (unless you go into management)
I worked for Lockheed and several other major corporations as an engineer, and the standard practice is to hire 'em, and fire 'em. One Christmas they corralled us all together and told us they were going to lay off 110 engineers. Being the youngest in the group, I thought I would be going. But no! It was the guys with 20+ years that got the ax. Guys with kids in college, with mortgages, who'd been loyal to the company. I saw my future and got the hell out. I'm in IT now, and even though things have been a little rough since dot bomb, they worst year in IT is better than the best year in engineering.
Why go to school for fours years in a very difficult subject only to get treated like cattle? Engineers make the world run, they make things that absolutely cannot ever break, live up to impossible standards, spend years in training, and get absolutely not gratitude whatsoever in return, either in salary or respect. I think its time they unionize.
I think this has become clear even to the kids. I remember my wife was offered a full ride to a very prestigious school for engineering. She went to a couple of companies in high school to see what engineers do, and turned it down. She paid to go to a state school, got a degree in communications, and is much happier for it.
When engineers start getting treated better, then more people will do it.
For those of you on the left side of the plane, please disregard the flames coming out of the engine. This only a slight malfunction and in no way presents any real danger to people in Tahiti. We have plenty of power to compensate for the engine that just fell off, as those of you on the right side may have just noticed. I assure you that this is still, in fact, a plane, and that it is currently in the air, so technically we are still flying. Furthermore, I want to assure all of you that we will continue to fly until we reach the ground, and I guarantee that all of you will reach the ground. So, technically, there is no problem at all folks. No problem at all. Everything is fine. Just fine.
Again, my name is Darl McBride, and thank you for flying SCO.
It's been two years and still there is no self-hosting OpenSolaris distribution. Again, there is no self-hosting OpenSolaris distribution. Again, there is yet to be ANY self-hosting OpenSolaris distribution. Not Nexenta, not Belenix, not Schillix, and sorry but Solaris Express is not open nor freely redistributable.
Source or no source, if that damn thing can't even be made to be self-hosting, and the resulting product freely-redistributable, then it can't even be compared with Linux, much less overtake it. Enough with the smoke and mirrors already
I fell for this hype two years ago when all the rage about Solaris 10 came out. Here's the deal: ZFS - great. DTrace - amazing. The Solaris kernel - truly exceptional. The userland, installer, package system, and general feel of the OS - horrendously bad... so awful that it sent all of us who tried it screaming back to Linux and BSD. And they are still going to stick with that awful package system -- even after Nexenta has done all the work to get Apt working, even after hiring Ian Murdock. And that's the amazing thing: Nexenta is a shining example of a budding community that has filled in almost every glaring gap that Solaris was lacking and rather than gobble it up, Sun has basically patted it on the head like a good little wannabe and marched right on by drunk in its typical, massive, NIH syndrome.
Not a chance. Keep the press releases coming, hire all the Linux people you want, but at the end of the day, I have at least two choices for a self-hosting, community-driven operating system with package systems, installers, and userlands that work now, not in years to come.
And Sun, please stop with the "we're gonna beat Linux" crap. Haven't you learned by now that that doesn't help you. The whole "us verses them" mentality has no place in the community, and just makes you look like an ass. Linux earned its place. Earn yours, with action, not press releases.
it's just a pain in the ass after using almost any Linux distro.
I know that's a troll comment. I don't mean to be a troll about it, but if I were to distill all of my experiences with Solaris, it really would come down to that. More than anything, it seems like Solaris consistently excelled at wasting large amounts of my time. And I don't think it was just because I was unfamiliar with it. I became familiar with it. I tweaked it until it almost looked and felt like Linux, and still it was a pain to use.
While ZFS is incredible, and DTrace amazing, there are so many other aspects of the system that are just horrendous. The package system, the userland, the complete (and intentional) lack of virtual terminals, the installer (this is a whole new world of pain). The installer is singularly the worst computing experience I've ever had, bar none. And don't lecture me about jumpstart. New users don't use jumpstart, they use that crappy-ass installer that is enough to put even the most devoted fanboy off Solaris. And this really tells the story about Solaris. While it has an amazing kernel, Sun has just completely ignored the critical features needed to recruit and retain new users.
Solaris needs community support, yet Solaris, even OpenSolaris, is still not self-hosting. Solaris is not open source in the way Linux is. The source is there, but for all practical purposes it is useless. There is no official OpenSolaris distro. You have to install Solaris Express, muck around with things, and then if you are lucky enough to get things compiled, you have this kind of hybrid, non-redistributable thing that sits in a legal gray area. Furthermore, even if you get this far, your "open" system is liable to be completely out of date in a month because there is no way to incrementally upgrade the kernel source. On "flag" days, you have to use a utility which is little more than a "this works in most cases but don't use it production" hack to install the new source and utilities. So to even get a system with the kernel source, you will not be able to reliably keep it up to date, or have any assurance that is even stable. Contrast this with having the source to the stable Linux kernel as a standard part of the OS. Forget the idea of having anything like 'make menuconfig.' So in many respects, Solaris being "open" is more marketing than practical reality.
And while there is Nexenta (Ubuntu with a Solaris kernel), which is an amazing feat, and already about as close to a Linux system running a Solaris kernel that you can get, they receive almost no support from Sun. As wonderful as Nexenta is, it still suffers from the fact that not all of OpenSolaris being completely open. That last I looked, it had no man pages, b/c Sun had not released them. They had to hack libm, as it was not available for a long time, and they had to hack their libc because Solaris' libc had strange dependencies on their (long broken) ksh implementation, which was not released as well. Furthermore, it, like every OpenSolaris distro is not self-hosting. And, rather than just embracing Nexenta's fabulous work in this area, Sun massive NIH complex demands that it make Solaris more Linux-like things it's own way.
There is little doubt in my mind that the Solaris kernel is one of the finest operating system kernels in existence, and is far superior to the Linux kernel. Sun's problem is that not only is everything surrounding that kernel stagnant, but that it really hasn't done the basic things needed to build a real community. Until OpenSolaris really is an open Solaris, with a stable, compilable kernel which can be incrementally upgraded and maintained by users, Solaris simply will not gain the support of the open source community. And that is what really matters today. I can Google "Ubuntu kidney" and find some informative post on how somebody configured Edgy to run a dialysis machine. That is, if I have a problem, I can get answers. Community support is more powerful than Sun support. I know, I've used both. And withou
This has got to be the most backward-ass, half-baked, clueless assessment of why to use a product that I've ever seen --- "Please give me just one thing, because I am incapable of choosing from many other great alternatives. I'm incapable of developing my product so that it will work. I must be shown the way."
Let me give you a good reason why one and only one choice sucks -- when your one and only choice sucks. Visual Studio 2005 was, and in large part still is, a complete disaster. Don't believe me? Google 'Microsoft Visual Studio 2005 sucks'. Follow the first link, and read with horror how even the most ardent MS dev fan boys had their asses set ablaze by their wonderful lack of choice:
I've worked with Visual Studio for years (not weeks), starting with Visual C++ 1.52. I've seen it rise and fall, and let me tell you, it's falling now. Many experienced developers are still clinging to VS 6 circa 98 because everything since just plain sucks. That's what you get when have no choice -- no choice. Ask the ISV's that developed with Visual J++ how they felt when MS and Sun got into it and MS, throwing a hissy fit, just *drops* the product. Ask people who have developed millions of lines of code in Visual Basic how they felt when MS abandoned them and EOL'd Visual Basic. There's lack of choice for you.
This guy has one week working with a MS product, and now he's telling the rest of us the error in our ways. Why is this clueless drivel even on slashdot? I understand the right to post whatever your like on your blog, but you call this slashdot worthy?
Too much choice. You've got to be joking. Is that the best argument for using MS products?
I got on the Lua bandwagon last year, and for what it is designed to do, it is quite good.
Before I start cheerleading, some comments on the book. Beginning Lua Programming (BLP) is a very good book on the subject, especially for people new to Lua or perhaps even to programming. I own this book as well as the "Programming in Lua" (PIL) book by Roberto Ierusalimschy, the latter of which I think has some of the best technical writing I've seen in a computer book and is on par with the likes of the late W. Richard Stevens. The two books are both very good, but I think serve slightly different audiences. BLP is a comprehensive book filled with gradual, well thought-out examples that can pretty much lead anyone through Lua regardless of programming experience, and help them to learn it well. The PIL book is written for more experienced/professional programmers who don't need as much hand holding.
Despite being a small language, Lua sports some serious language features, and you will be sorry not to learn them all. This is where the two books differ I think. PIL assumes at least a nodding acquaintance with concepts such as closures, iterators, generators, and more advanced concepts such as its coroutines (which are a somewhat extended form of blocks in Ruby). If you are not familiar with these, BLP will ensure that you come away with a full understanding through a battery of examples and patient prose.
That said, people who are already familiar with the above (through prior experience with languages such as Ruby or Python, for example) might find BLP a little slow. This is where PIL really shines. PIL is an extremely well-written book by the Lua's principle architect which provides a full treatment of the language in an economy of pages. It is hard to believe when reading the text that English is the author's second language. The examples are all very good, well chosen, but not excessive. The book moves at a steady pace and covers the same ground in much less space than BLP. One place where PIL really shines is on covering the C API and C programming aspects of Lua. In this respect, it is clearly superior to BLP. But again, you have to be a well-grounded C/C++ developer to appreciate this.
As I said, I own them both. I read PIL first which got me embedding Lua in C applications in no time, and then went back and read BLP to round out and sharpen my skills in the language proper. I think the books can complement each other in this way. They are both above average books.
And now for those who have not used Lua and wonder why they would want to. Lua is a tiny, standalone language/interpreter that can be put almost anywhere. It is an ideal way to extend your application as the entire language and libraries fit into a single library under 200Kb in size. It is written in ANSI C and is extremely portable. In my opinion, Lua is not a language you would choose instead of Ruby or Python (unless of course you were already very proficient at it), as I think those languages have much more to offer in either syntactical elegance or extensive standard libraries. Lua's function as a language is very similar to SQLite's function as a database. It's a powerful little multitool you can use to dramatically enhance and extend your applications with very little effort. A feature that might take 3,000 lines of C might take 10 lines of Lua. My experience with it so far has been very positive, and it has pretty much become a staple in my programming repertoire. It is the ultimate compact domain specific language.
This reminds me of the quote by Brander Matthews:
"A gentleman need not know Latin, but he should at least have forgotten it."
As this was said of the educated a century ago, the same might be said of assembly language for programmers today. There are benefits outside of direct application that prove worthy in other endeavors.
Subject : [TAX GEEK] Announce Date: April 18, 2007
We are pleased to announce our latest update of Tax Geek, which fixes a critical off-by-one error in the previous release, which could in some cases lead to (severe) inaccuracies. Please update your current version ASAP. As always, if you find any additional errors, please submit bug reports (and preferably patches) on Source Forge, and Joe will look into when he gets home from class.
Linux may not be just a programmer's OS, but the Ubuntu flavor of Linux, IMO, isn't a very good programmer's OS at all. I think it crossed that fine line between control and ease of use.
That's just plain silly. I've been programming professionally for ten years and in my experience, Ubuntu is exactly what you want for a developer's workstation. It has practically zero maintenance, installs right out of the box with little to no configuration, almost every major software package is available in binary form, being little more than an apt-get away. Beryl, OpenOffice, Apache, PostgreSQL, MySQL, Rails, you name it -- Ubuntu has the most up to date builds of everything, wonderfully configured, and ready to install. Do you need the GCC toolchain and related developers tools? apt-get build-essential. Done. Do you need various multimedia or Win32 codecs not in the main repos? apt-get easyb-ubuntu or automatix and with a click-and-drool interface that even a Windows user would love, a few mouse clicks will fetch and install it all, plus some stuff you hadn't thought of or knew existed.
You have over 12,000 available software packages compiled for Ubuntu, in the official repository alone, not to mention all the others. I came from Gentoo, and Ubuntu is infinitely better because I can put my effort into building *my* software, rather than everybody else's. That is the point of programming, right? Working on your software, not spending days compiling your system from scratch.
Yes, there are some dependency issues with some packages, in that they may link with other libs you may not want (e.g. amarok brings in MySQL client lib), but this is true with *all* distros. You are at the mercy of the package maintainer. If you don't agree, you have to compile from scratch -- as with any other distro. But typically with a workstation, who cares if you bring in other libs? You've got tons of disk space and the goal is comfort and ease. Perhaps on a server you may mince over deps, but that's another story.
And -- getting on my soapbox I just have to add it -- since Linux/UNIX/OSS is so incredible, once I have my system exactly as I want it, I drop to single user mode, mount a USB drive and to do a dump of my root filesystem, making an exact backup image of my system. From that point on, I will *never* have to reinstall from scratch no matter what happens. If my drive hoses, I simply boot from a live CD of any distro, create a root file system, restore from my backup image, update grub, and reboot. Right back to normal. No online activation or phoning home, no install keys, nothing -- just rewrite the system image to disk. What would take three hours in reinstalling Windows (reinstalling the OS, activiation, and all installing other software) takes under 5 minutes with Linux. Of course, if I *buy* more software with Windows (e.g.Ghost), I could follow a similar process, but it still takes longer, and is still more of a pain in the ass. I've done it -- I know. My solution now is to run Windows in Linux using VMWare player (free). Now, my Windows partition is actually sitting inside a Linux file system, which I back up to USB using the aforementioned process. Windows backups/reinstalls are now as easy as Linux (thumbs nose at M$).
Sounds like he was using Gentoo. We once had a single emerge of lsusb swap out the system's glibc, apache, Samba, PostgreSQL, and just about everything else. We had to restore the entire system from backup. To be fair, there is no such thing as piecemeal dependency management beyond a certain point, before you've just got to swap out the whole userland. It doesn't matter if you are using RPM, portage, or apt, if your system gets far enough out of date, it's only a matter of time until this happens.
Moving to Ubuntu probably is the best solution however, in that they actively support better QC and more current builds. Plus the sheer number of users is a plus as well. I didn't want to move to Ubuntu at first, but after a while it just made sense. You can google "ubuntu kidney transplant" nowadays and find someone out there whose done it with Dapper or Edgy. Better install out of the box, more current software, lots of users, good documentation, good package system (apt/dpkg may not be perfect, but it is an order of magnitude better than anything else). Ubuntu is just plain easier. Hell, it's not a bad server either, although I prefer/use BSD for that.
Microsoft has the best virtual machine with.NET, the best development tool with Visual Studio and the best access to developers with their MSDN programs. What an unbelievably biased and unsupportable statement. Best virtual machine? What? Best development tool? Ask these people how great they think Visual Studio really is. And these are the people that actually use MS development products.
Is this journalism, or yet more thinly veiled MS fanboy talk? I love how everything Microsoft always has to be cast in the XYZ-Killer, or storm brewing, or some other ominous "better watch out" metaphor. Isn't this the same kind of talk that was used to describe Vista, before virtually every ground-breaking feature was a no-show, and yet again we're just going to see a pretty changed-up GUI.
When MS can just create software that works, I'll be more inclined to actually give the "killer" statements some consideration. Until then, I'll file this crap away with WinFS, and Zune "the great iPod killer."
I know my life is much better now that I can see crisp, informative error messages explaining why IE is not working. Now, if they could just do that for Windows. Perhaps Vista.
Slashdot Mirror
...we will have done just that - positioned Sun at the center of the web What part of LAMP stands for Sun? And merely buying the 'M' doesn't count.There is no turn around in Dell. And there was better Linux/BSD support for Linux *before* Dell supported Linux.
We've been a Dell customer for 10 years. The first five years were pretty good. The old PowerEdge servers were pretty nice. But I've never met a Perc card I liked.
What is a Perc card? Well, it is generic term for the card that Dell happens to be rebranding at the moment.
What is rebranding? Well, take a perhaps decent card, screw with it for the sake of vendor lock-in, so that the original vendor's drivers and utilities no longer work with it, degrade its performance beyond measure without any hope of cure, and you have a Dell rebrand.
The new MD1000, while pretty nice arrays, are only officially supported if you use a Perc 5e card. This card is a true Dell rebrand of an LSI MegraRAID. It's a thousand dollar turd. There are no decent utilities (in fact all LSI megaraid utils will NOT work). They even took away some of the simple command line utilities for Linux like lsiutil, and replaced it with a huge, bloated Java application. Can you imagine that piece of bloatware as your only means of talking to your drive array? And FreeBSD support, while the FreeBSD community has made a valiant effort trying to support this piece of shit, still is not worth it. It's lackluster performance at best.
But here's the real kicker about the Perc 5e: It does not allow direct pass through. Yeah, that's right, if you get fed up with it's lame RAID 10 performance (and it is lame we've confirmed that software RAID on Linux and FreeBSD is significantly faster), you can't directly reach your drives. So if you want to implement software RAID, you have to configure independent RAID0 virtual disks in the Perc 5e BIOS. Why? so Dell can lock you in to their proprietary, rebranded, turd card. This is a wonderful new feature unique to the Perc 5 that wasn't there with Perc 4.
But, as we found out, the Perc 4 was an even bigger piece of crap. It was a rebranded Adaptec that had over a year-long history of locking up under heavy loads on Linux systems. They were selling this so-called Linux supported PowerEdge 2650 (boat anchor) for over a year knowing that it locked up under stress:
https://bugzilla.redhat.com/show_bug.cgi?id=92129
We got burned by this and it took us weeks to resolve (went through 3 Perc 4 cards with same result, and ended up putting in a LSI card).
If you need anything with real RAID support, run from Dell. We've been pretty pleased with the MD1000, but that Perc 5 card sums up Dell. Buy a 3ware or Areca instead. Who cares if Dell won't support the MD1000 if you use it, they don't support it anyway. Of all our calls to Dell enterprise support, we've talked to one person who seems to have half of a clue.
There is no such thing as Dell support. And their lame attempt at vendor lock-in only makes it worse. And that is why Dell is drowning.
We've moved to HP. And we've been really happy with our HP servers. Is HP support any better? I can't say. But we don't have near the vendor lock-in and rebranded voodoo we get from Dell.
I have a masters in chemical engineering. I come from a family of engineers: my dad's an engineer, his dad was one, and my father in law is one -- altogether we cover electrical, civil, chemical, and industrial.
Here's my take on why so few Americans are going into engineering:
I worked for Lockheed and several other major corporations as an engineer, and the standard practice is to hire 'em, and fire 'em. One Christmas they corralled us all together and told us they were going to lay off 110 engineers. Being the youngest in the group, I thought I would be going. But no! It was the guys with 20+ years that got the ax. Guys with kids in college, with mortgages, who'd been loyal to the company. I saw my future and got the hell out. I'm in IT now, and even though things have been a little rough since dot bomb, they worst year in IT is better than the best year in engineering.
Why go to school for fours years in a very difficult subject only to get treated like cattle? Engineers make the world run, they make things that absolutely cannot ever break, live up to impossible standards, spend years in training, and get absolutely not gratitude whatsoever in return, either in salary or respect. I think its time they unionize.
I think this has become clear even to the kids. I remember my wife was offered a full ride to a very prestigious school for engineering. She went to a couple of companies in high school to see what engineers do, and turned it down. She paid to go to a state school, got a degree in communications, and is much happier for it.
When engineers start getting treated better, then more people will do it.Hello, this is your captain speaking.
For those of you on the left side of the plane, please disregard the flames coming out of the engine. This only a slight malfunction and in no way presents any real danger to people in Tahiti. We have plenty of power to compensate for the engine that just fell off, as those of you on the right side may have just noticed. I assure you that this is still, in fact, a plane, and that it is currently in the air, so technically we are still flying. Furthermore, I want to assure all of you that we will continue to fly until we reach the ground, and I guarantee that all of you will reach the ground. So, technically, there is no problem at all folks. No problem at all. Everything is fine. Just fine.
Again, my name is Darl McBride, and thank you for flying SCO.
It's been two years and still there is no self-hosting OpenSolaris distribution. Again, there is no self-hosting OpenSolaris distribution. Again, there is yet to be ANY self-hosting OpenSolaris distribution. Not Nexenta, not Belenix, not Schillix, and sorry but Solaris Express is not open nor freely redistributable.
Source or no source, if that damn thing can't even be made to be self-hosting, and the resulting product freely-redistributable, then it can't even be compared with Linux, much less overtake it. Enough with the smoke and mirrors already
I fell for this hype two years ago when all the rage about Solaris 10 came out. Here's the deal: ZFS - great. DTrace - amazing. The Solaris kernel - truly exceptional. The userland, installer, package system, and general feel of the OS - horrendously bad ... so awful that it sent all of us who tried it screaming back to Linux and BSD. And they are still going to stick with that awful package system -- even after Nexenta has done all the work to get Apt working, even after hiring Ian Murdock. And that's the amazing thing: Nexenta is a shining example of a budding community that has filled in almost every glaring gap that Solaris was lacking and rather than gobble it up, Sun has basically patted it on the head like a good little wannabe and marched right on by drunk in its typical, massive, NIH syndrome.
Not a chance. Keep the press releases coming, hire all the Linux people you want, but at the end of the day, I have at least two choices for a self-hosting, community-driven operating system with package systems, installers, and userlands that work now, not in years to come.
And Sun, please stop with the "we're gonna beat Linux" crap. Haven't you learned by now that that doesn't help you. The whole "us verses them" mentality has no place in the community, and just makes you look like an ass. Linux earned its place. Earn yours, with action, not press releases.
it's just a pain in the ass after using almost any Linux distro.
I know that's a troll comment. I don't mean to be a troll about it, but if I were to distill all of my experiences with Solaris, it really would come down to that. More than anything, it seems like Solaris consistently excelled at wasting large amounts of my time. And I don't think it was just because I was unfamiliar with it. I became familiar with it. I tweaked it until it almost looked and felt like Linux, and still it was a pain to use.
While ZFS is incredible, and DTrace amazing, there are so many other aspects of the system that are just horrendous. The package system, the userland, the complete (and intentional) lack of virtual terminals, the installer (this is a whole new world of pain). The installer is singularly the worst computing experience I've ever had, bar none. And don't lecture me about jumpstart. New users don't use jumpstart, they use that crappy-ass installer that is enough to put even the most devoted fanboy off Solaris. And this really tells the story about Solaris. While it has an amazing kernel, Sun has just completely ignored the critical features needed to recruit and retain new users.
Solaris needs community support, yet Solaris, even OpenSolaris, is still not self-hosting. Solaris is not open source in the way Linux is. The source is there, but for all practical purposes it is useless. There is no official OpenSolaris distro. You have to install Solaris Express, muck around with things, and then if you are lucky enough to get things compiled, you have this kind of hybrid, non-redistributable thing that sits in a legal gray area. Furthermore, even if you get this far, your "open" system is liable to be completely out of date in a month because there is no way to incrementally upgrade the kernel source. On "flag" days, you have to use a utility which is little more than a "this works in most cases but don't use it production" hack to install the new source and utilities. So to even get a system with the kernel source, you will not be able to reliably keep it up to date, or have any assurance that is even stable. Contrast this with having the source to the stable Linux kernel as a standard part of the OS. Forget the idea of having anything like 'make menuconfig.' So in many respects, Solaris being "open" is more marketing than practical reality.
And while there is Nexenta (Ubuntu with a Solaris kernel), which is an amazing feat, and already about as close to a Linux system running a Solaris kernel that you can get, they receive almost no support from Sun. As wonderful as Nexenta is, it still suffers from the fact that not all of OpenSolaris being completely open. That last I looked, it had no man pages, b/c Sun had not released them. They had to hack libm, as it was not available for a long time, and they had to hack their libc because Solaris' libc had strange dependencies on their (long broken) ksh implementation, which was not released as well. Furthermore, it, like every OpenSolaris distro is not self-hosting. And, rather than just embracing Nexenta's fabulous work in this area, Sun massive NIH complex demands that it make Solaris more Linux-like things it's own way.
There is little doubt in my mind that the Solaris kernel is one of the finest operating system kernels in existence, and is far superior to the Linux kernel. Sun's problem is that not only is everything surrounding that kernel stagnant, but that it really hasn't done the basic things needed to build a real community. Until OpenSolaris really is an open Solaris, with a stable, compilable kernel which can be incrementally upgraded and maintained by users, Solaris simply will not gain the support of the open source community. And that is what really matters today. I can Google "Ubuntu kidney" and find some informative post on how somebody configured Edgy to run a dialysis machine. That is, if I have a problem, I can get answers. Community support is more powerful than Sun support. I know, I've used both. And withou
This has got to be the most backward-ass, half-baked, clueless assessment of why to use a product that I've ever seen --- "Please give me just one thing, because I am incapable of choosing from many other great alternatives. I'm incapable of developing my product so that it will work. I must be shown the way."
Let me give you a good reason why one and only one choice sucks -- when your one and only choice sucks. Visual Studio 2005 was, and in large part still is, a complete disaster. Don't believe me? Google 'Microsoft Visual Studio 2005 sucks'. Follow the first link, and read with horror how even the most ardent MS dev fan boys had their asses set ablaze by their wonderful lack of choice:
http://www.microsoftweblog.com/2005/11/05/problems -with-visual-studio-2005/
I've worked with Visual Studio for years (not weeks), starting with Visual C++ 1.52. I've seen it rise and fall, and let me tell you, it's falling now. Many experienced developers are still clinging to VS 6 circa 98 because everything since just plain sucks. That's what you get when have no choice -- no choice. Ask the ISV's that developed with Visual J++ how they felt when MS and Sun got into it and MS, throwing a hissy fit, just *drops* the product. Ask people who have developed millions of lines of code in Visual Basic how they felt when MS abandoned them and EOL'd Visual Basic. There's lack of choice for you.
This guy has one week working with a MS product, and now he's telling the rest of us the error in our ways. Why is this clueless drivel even on slashdot? I understand the right to post whatever your like on your blog, but you call this slashdot worthy?
Too much choice. You've got to be joking. Is that the best argument for using MS products?
I got on the Lua bandwagon last year, and for what it is designed to do, it is quite good.
Before I start cheerleading, some comments on the book. Beginning Lua Programming (BLP) is a very good book on the subject, especially for people new to Lua or perhaps even to programming. I own this book as well as the "Programming in Lua" (PIL) book by Roberto Ierusalimschy, the latter of which I think has some of the best technical writing I've seen in a computer book and is on par with the likes of the late W. Richard Stevens. The two books are both very good, but I think serve slightly different audiences. BLP is a comprehensive book filled with gradual, well thought-out examples that can pretty much lead anyone through Lua regardless of programming experience, and help them to learn it well. The PIL book is written for more experienced/professional programmers who don't need as much hand holding.
Despite being a small language, Lua sports some serious language features, and you will be sorry not to learn them all. This is where the two books differ I think. PIL assumes at least a nodding acquaintance with concepts such as closures, iterators, generators, and more advanced concepts such as its coroutines (which are a somewhat extended form of blocks in Ruby). If you are not familiar with these, BLP will ensure that you come away with a full understanding through a battery of examples and patient prose.
That said, people who are already familiar with the above (through prior experience with languages such as Ruby or Python, for example) might find BLP a little slow. This is where PIL really shines. PIL is an extremely well-written book by the Lua's principle architect which provides a full treatment of the language in an economy of pages. It is hard to believe when reading the text that English is the author's second language. The examples are all very good, well chosen, but not excessive. The book moves at a steady pace and covers the same ground in much less space than BLP. One place where PIL really shines is on covering the C API and C programming aspects of Lua. In this respect, it is clearly superior to BLP. But again, you have to be a well-grounded C/C++ developer to appreciate this.
As I said, I own them both. I read PIL first which got me embedding Lua in C applications in no time, and then went back and read BLP to round out and sharpen my skills in the language proper. I think the books can complement each other in this way. They are both above average books.
And now for those who have not used Lua and wonder why they would want to. Lua is a tiny, standalone language/interpreter that can be put almost anywhere. It is an ideal way to extend your application as the entire language and libraries fit into a single library under 200Kb in size. It is written in ANSI C and is extremely portable. In my opinion, Lua is not a language you would choose instead of Ruby or Python (unless of course you were already very proficient at it), as I think those languages have much more to offer in either syntactical elegance or extensive standard libraries. Lua's function as a language is very similar to SQLite's function as a database. It's a powerful little multitool you can use to dramatically enhance and extend your applications with very little effort. A feature that might take 3,000 lines of C might take 10 lines of Lua. My experience with it so far has been very positive, and it has pretty much become a staple in my programming repertoire. It is the ultimate compact domain specific language.
I compiled and installed the plugin and every time I try it, get the following:
Surprise, Windows Listed as Most Secure OS ... just don't move the mouse.
As this was said of the educated a century ago, the same might be said of assembly language for programmers today. There are benefits outside of direct application that prove worthy in other endeavors.
That's just plain silly. I've been programming professionally for ten years and in my experience, Ubuntu is exactly what you want for a developer's workstation. It has practically zero maintenance, installs right out of the box with little to no configuration, almost every major software package is available in binary form, being little more than an apt-get away. Beryl, OpenOffice, Apache, PostgreSQL, MySQL, Rails, you name it -- Ubuntu has the most up to date builds of everything, wonderfully configured, and ready to install. Do you need the GCC toolchain and related developers tools? apt-get build-essential. Done. Do you need various multimedia or Win32 codecs not in the main repos? apt-get easyb-ubuntu or automatix and with a click-and-drool interface that even a Windows user would love, a few mouse clicks will fetch and install it all, plus some stuff you hadn't thought of or knew existed.
You have over 12,000 available software packages compiled for Ubuntu, in the official repository alone, not to mention all the others. I came from Gentoo, and Ubuntu is infinitely better because I can put my effort into building *my* software, rather than everybody else's. That is the point of programming, right? Working on your software, not spending days compiling your system from scratch.
Yes, there are some dependency issues with some packages, in that they may link with other libs you may not want (e.g. amarok brings in MySQL client lib), but this is true with *all* distros. You are at the mercy of the package maintainer. If you don't agree, you have to compile from scratch -- as with any other distro. But typically with a workstation, who cares if you bring in other libs? You've got tons of disk space and the goal is comfort and ease. Perhaps on a server you may mince over deps, but that's another story.
And -- getting on my soapbox I just have to add it -- since Linux/UNIX/OSS is so incredible, once I have my system exactly as I want it, I drop to single user mode, mount a USB drive and to do a dump of my root filesystem, making an exact backup image of my system. From that point on, I will *never* have to reinstall from scratch no matter what happens. If my drive hoses, I simply boot from a live CD of any distro, create a root file system, restore from my backup image, update grub, and reboot. Right back to normal. No online activation or phoning home, no install keys, nothing -- just rewrite the system image to disk. What would take three hours in reinstalling Windows (reinstalling the OS, activiation, and all installing other software) takes under 5 minutes with Linux. Of course, if I *buy* more software with Windows (e.g.Ghost), I could follow a similar process, but it still takes longer, and is still more of a pain in the ass. I've done it -- I know. My solution now is to run Windows in Linux using VMWare player (free). Now, my Windows partition is actually sitting inside a Linux file system, which I back up to USB using the aforementioned process. Windows backups/reinstalls are now as easy as Linux (thumbs nose at M$).
Sounds like he was using Gentoo. We once had a single emerge of lsusb swap out the system's glibc, apache, Samba, PostgreSQL, and just about everything else. We had to restore the entire system from backup. To be fair, there is no such thing as piecemeal dependency management beyond a certain point, before you've just got to swap out the whole userland. It doesn't matter if you are using RPM, portage, or apt, if your system gets far enough out of date, it's only a matter of time until this happens.
Moving to Ubuntu probably is the best solution however, in that they actively support better QC and more current builds. Plus the sheer number of users is a plus as well. I didn't want to move to Ubuntu at first, but after a while it just made sense. You can google "ubuntu kidney transplant" nowadays and find someone out there whose done it with Dapper or Edgy. Better install out of the box, more current software, lots of users, good documentation, good package system (apt/dpkg may not be perfect, but it is an order of magnitude better than anything else). Ubuntu is just plain easier. Hell, it's not a bad server either, although I prefer/use BSD for that.
Is this journalism, or yet more thinly veiled MS fanboy talk? I love how everything Microsoft always has to be cast in the XYZ-Killer, or storm brewing, or some other ominous "better watch out" metaphor. Isn't this the same kind of talk that was used to describe Vista, before virtually every ground-breaking feature was a no-show, and yet again we're just going to see a pretty changed-up GUI.
When MS can just create software that works, I'll be more inclined to actually give the "killer" statements some consideration. Until then, I'll file this crap away with WinFS, and Zune "the great iPod killer."
I know my life is much better now that I can see crisp, informative error messages explaining why IE is not working. Now, if they could just do that for Windows. Perhaps Vista.