To reiterate, that one server passes every hardware diagnostic, runs fine on a 2.4.x kernel (uptime in weeks), and has identical configuration to another server that's running 2.6 just fine.
You fetch the relevant bit of file via HTTP? It's really just like implementing disk caching.
Of course, there's the pathological worst case where the ID3 tag is ID3v1 or ID3v2.4, i.e. at the end of the file, and the HTTP server doesn't support HTTP/1.1 byte range requests. In that case, you fetch the entire file. But that's no worse than not having the middle caching layer at all, and it's hard to see how you could do better in that case.
At the risk of pointing out the very obvious, Interactive Fiction will work fine for blind users, via a screen reader. There are many excellent freeware IF games, comparable to Infocom's finest.
I was running Atari TOS. AST's microkernel design ran on my Atari ST and gave me a working Unix-like system. Linus's hack only ran on the 386, and was useless to me. Arguing that it was superior when it only ran on a machine I didn't have was pointless.
Not only did they make it only work on a Mac-only server... they also didn't bother to produce a demo version. No way was I about to drop $50 on a game without even being able to see it running first.
How about using the existing C id3v2 library, but feeding it a UNIX socket/pipe instead of an actual file or stream? Then at the other end of the pipe you put some code which does just-in-time fetching of appropriate chunks of the MP3 at the other end of the network connection.
Of course, this relies on the library seeking in a sensible way, and you might have to hack it to use seek to determine file size rather than fstat.
A lot of people here have said that they think it's easy to create original software inventions. I would ask them if they've ever done it.
Well, the XOR graphics cursor is a patented "software invention", and I independently invented that when I was 12 years old.
I also independently invented downloadable software delivered by following links on a hypertext page, a couple of years after that, around 1983. Another patented "software invention".
I'm not joking about this. Of course, at the time software patents didn't exist; if I'd thought to file for patents on "my" inventions, I could have raked in a fortune.
So no, my perspective is that creating original software inventions is something a teenage kid can do. Ideas are easy, and creating a "software invention" is pretty much a matter of having an idea and writing it down in special notation.
I do believe there are a few algorithms which are sufficiently devious and clever that they might be deserving of patent protection. RSA is one example, but ironically it's also a great example of horrific abuse of the patent system.
In my view, the basic goals of CM are: (a) know what you shipped, (b) know when you shipped it, (c) deliver releases on time as predicted, (d) have product functionality decisions actually under the control of product management, and (e) drive product functionality from actual customer requirements.
Set out like that, it seems blatantly obvious that that's the way things should be, but when you start trying to make things work that way you'll find that the required changes to working practices are controversial.
Lots of developers will hate you. Developers like to be able to go fix any problem they happen to notice without telling anyone, and they hate documenting things. They don't test adequately enough to even ensure that their fixes work, far less prevent regressions. (Yes, that's gross generalizing and unfair to many, but I'm also a developer.)
The QA and documentation teams will love you, for exactly the same reasons.
If the current situation is a mess, which it probably is if they're desperate enough to offer the job to someone with no experience, it'll take at least a year to straighten it out.
In fact, before accepting the position, make sure it isn't a Poison Chalice Project.
Expect massive political battles, flamewars, perhaps even sabotage by primadonna developers.
Make friends with other clueful people who've worked on CM and process improvements. Run ideas past them, discuss problems with them, and so on. They don't necessarily have to be people at the same company.
Also, bear in mind that if you manage to fix things, and development suddenly start shipping reasonably bug-free releases on time, you are unlikely to get any of the credit. The prima donna programmers and project managers will get the glory.
OS X 10.3 made major improvements in reliability. 10.2 would regularly corrupt the filesystem if powered off hard or if the kernel crashed, but I've yet to have 10.3 crashes result in any kind of error (as verified by Disk Utility and Disk Warrior).
Similarly, I have three Linux servers running with the root filesystem on ReiserFS, all running recent 2.4 kernels or 2.6. They've all been powered off in the middle of writing to disk, and none of them has suffered any kind of data loss or corruption to date.
I really think ReiserFS has an unjustified bad rep. Maybe it was flaky once, but not these days.
"I bought this product at a mall and it turned out to be a TOTAL RIP-OFF. I am NEVER going to buy ANYTHING at a mall EVER AGAIN!"
If people ranted like that, you'd write them off as whackos. For some reason, when they say the same thing about eBay rather than a mall, we pay credence to them, even though everyone's expectations of eBay are lower than their expectations of a mall.
Oh, give me a break. I walk around Kendall at 10-11pm on a regular basis, you make it sound like some kind of post-apocalyptic hellhole.
There's a steady stream of people walking to Central Square, the Kendall Cinema, and so on. A block or two away from the Stata Center you'll find an Italian restaurant, Legal Seafood, a big Marriott Hotel, and so on, all regularly full of people late in the evening.
Am I the only person still having major problems with the Adaptec 7xxx drivers? They broke in 2.6 and they still aren't working on one of my machines. They work on all the others, though, so there's something subtly not quite right...
Re:you can't?
on
GPS for GBA
·
· Score: 3, Insightful
If I could have a reliable GPS unit that was attached to my PDA and it still functioned as well as a handheld one I would love it.
Slashdot Mirror
I always felt that the AGR was a much better idea than the PWR, because in the event of accident convection would keep the core cool.
That article was a crock, though. There are plenty of Prius users who get way over the rated mileage. It depends a lot on how you drive the thing.
We rented a Prius for a vacation in San Francisco. We got 55mpg driving to Monterey to see the otters, and 70mpg in the city.
Uh, yeah, perhaps you have some suggestions?
To reiterate, that one server passes every hardware diagnostic, runs fine on a 2.4.x kernel (uptime in weeks), and has identical configuration to another server that's running 2.6 just fine.
You fetch the relevant bit of file via HTTP? It's really just like implementing disk caching.
Of course, there's the pathological worst case where the ID3 tag is ID3v1 or ID3v2.4, i.e. at the end of the file, and the HTTP server doesn't support HTTP/1.1 byte range requests. In that case, you fetch the entire file. But that's no worse than not having the middle caching layer at all, and it's hard to see how you could do better in that case.
At the risk of pointing out the very obvious, Interactive Fiction will work fine for blind users, via a screen reader. There are many excellent freeware IF games, comparable to Infocom's finest.
"We" weren't all running DOS, though.
I was running Atari TOS. AST's microkernel design ran on my Atari ST and gave me a working Unix-like system. Linus's hack only ran on the 386, and was useless to me. Arguing that it was superior when it only ran on a machine I didn't have was pointless.
Not only did they make it only work on a Mac-only server... they also didn't bother to produce a demo version. No way was I about to drop $50 on a game without even being able to see it running first.
How about using the existing C id3v2 library, but feeding it a UNIX socket/pipe instead of an actual file or stream? Then at the other end of the pipe you put some code which does just-in-time fetching of appropriate chunks of the MP3 at the other end of the network connection.
Of course, this relies on the library seeking in a sensible way, and you might have to hack it to use seek to determine file size rather than fstat.
So when are they going to fix XVID and DivX to use MPEG-4 conformant file formats instead of AVI?
Well, the XOR graphics cursor is a patented "software invention", and I independently invented that when I was 12 years old.
I also independently invented downloadable software delivered by following links on a hypertext page, a couple of years after that, around 1983. Another patented "software invention".
I'm not joking about this. Of course, at the time software patents didn't exist; if I'd thought to file for patents on "my" inventions, I could have raked in a fortune.
So no, my perspective is that creating original software inventions is something a teenage kid can do. Ideas are easy, and creating a "software invention" is pretty much a matter of having an idea and writing it down in special notation.
I do believe there are a few algorithms which are sufficiently devious and clever that they might be deserving of patent protection. RSA is one example, but ironically it's also a great example of horrific abuse of the patent system.
In my view, the basic goals of CM are: (a) know what you shipped, (b) know when you shipped it, (c) deliver releases on time as predicted, (d) have product functionality decisions actually under the control of product management, and (e) drive product functionality from actual customer requirements.
Set out like that, it seems blatantly obvious that that's the way things should be, but when you start trying to make things work that way you'll find that the required changes to working practices are controversial.
Lots of developers will hate you. Developers like to be able to go fix any problem they happen to notice without telling anyone, and they hate documenting things. They don't test adequately enough to even ensure that their fixes work, far less prevent regressions. (Yes, that's gross generalizing and unfair to many, but I'm also a developer.)
The QA and documentation teams will love you, for exactly the same reasons.
If the current situation is a mess, which it probably is if they're desperate enough to offer the job to someone with no experience, it'll take at least a year to straighten it out.
In fact, before accepting the position, make sure it isn't a Poison Chalice Project.
Expect massive political battles, flamewars, perhaps even sabotage by primadonna developers.
Make friends with other clueful people who've worked on CM and process improvements. Run ideas past them, discuss problems with them, and so on. They don't necessarily have to be people at the same company.
Reading suggestions I haven't seen anyone else mention yet: "AntiPatterns",
"The Rational Unified Process".
Also, bear in mind that if you manage to fix things, and development suddenly start shipping reasonably bug-free releases on time, you are unlikely to get any of the credit. The prima donna programmers and project managers will get the glory.
Should fit right in with other open source toolkits like Eclipse/SWT, then.
Yeah, we got >70mpg in San Francisco, and over 50mpg on the highways, driving a rented first generation Prius for a week.
Actually, McDonalds have a "University of Hamburgerology" in Oak Brook IL, and it does issue degree certificates.
Not sure if they count as "bogus", though, and they're probably worth more than an MCSE.
OS X 10.3 made major improvements in reliability. 10.2 would regularly corrupt the filesystem if powered off hard or if the kernel crashed, but I've yet to have 10.3 crashes result in any kind of error (as verified by Disk Utility and Disk Warrior).
Similarly, I have three Linux servers running with the root filesystem on ReiserFS, all running recent 2.4 kernels or 2.6. They've all been powered off in the middle of writing to disk, and none of them has suffered any kind of data loss or corruption to date.
I really think ReiserFS has an unjustified bad rep. Maybe it was flaky once, but not these days.
Yeah, like Windows.
In all seriousness, if they sold VAIO systems without Windows, I'd have bought one.
It plays MP3 files by converting them to ATRAC on your PC.
Imagine how online culture would change if Microsoft went after end users, and split the profit 50/50 with the snitch.
"I bought this product at a mall and it turned out to be a TOTAL RIP-OFF. I am NEVER going to buy ANYTHING at a mall EVER AGAIN!"
If people ranted like that, you'd write them off as whackos. For some reason, when they say the same thing about eBay rather than a mall, we pay credence to them, even though everyone's expectations of eBay are lower than their expectations of a mall.
Coming soon at MIT...
The ENRON Business School
The Jayson Blair School of Journalism
The Timothy McVeigh School of Architecture
It's sad that universities are so desperate for cash that they're willing to whore themselves out to a bunch of crooks.
Kinda supports the idea that Bill only started giving away money to get positive PR for Microsoft, though, doesn't it?
(Remember the memo from his PR consultants that said so, that was leaked?)
Presumably the only reason Boston's now strict about how architecture looks, is that they learned their lesson from BU...
Oh, give me a break. I walk around Kendall at 10-11pm on a regular basis, you make it sound like some kind of post-apocalyptic hellhole.
There's a steady stream of people walking to Central Square, the Kendall Cinema, and so on. A block or two away from the Stata Center you'll find an Italian restaurant, Legal Seafood, a big Marriott Hotel, and so on, all regularly full of people late in the evening.
Am I the only person still having major problems with the Adaptec 7xxx drivers? They broke in 2.6 and they still aren't working on one of my machines. They work on all the others, though, so there's something subtly not quite right...
So buy a Garmin iQue.