Except more complex software is more error-prone, especially in the unpredictable failure-conditions you want to catch.
You really do not want to be vulnerable to lighting by having an antenna, and you probably want a highly error-tolerant processing architecture, to cope with the higher-levels of radiation. That may compromise performance and programming methodology. (I.E. read about single-bit-errors in normal PC-architectures)
Keep it simple, and make sure to add a tracking-beacon so they find the darn thing.
Salesperson: Look, our state-of-the art FDR records directly to the cloud! No data-box to find and retrieve, no fuss, and MapReduce there immediately after the accident to start the analysis. Airline manufacturer: Sounds great, I'll buy.
A couple of months pass, the first aircraft goes down...
Air Crash Investigator: It seems we had no 3g connection minutes before the crash, and the plane was pointing away from the satellite. We haven't got a clue. Salesperson: Oh well, better luck next time.
There's an argument for rugged well-tested simple designs here.
This doesn't seem to cover license-claims for implementations of H264, which most of the debate has been about lately? What happens to that, has it too been extended?
Every piece of specialized software will fail occasionally. Not talking highly productized software with a gazillion beta-testers, but the ERP integration layers and other type of software that has been more or less written for a single, or for a few customers.
No developer, and no amount of synthetical testing can ever cover all the possible angles in these highly complex systems the way the real world can. This is especially noticable for integration systems, highly dependent on external environment.
Some of this software is highly mission-critical, when it stops, business stops. In these cases, dev-access is probably the sound way to troubleshoot, and get things running smoothly again.
There are a few gaming companies producing Linux-versions, and selling on Steam. Aquaria, Osmos and World of Goo I know for sure, but I think also the Penumbra series, Savage, Lugaru, Overgrowth, Bridge Builder, Dark Horizons, Machinarium, Serious Sam, and probably some more are available on Steam (and certainly on Linux).
Especially, I think partnership with Cedega or CrossOver could have been one way to go about it.
Personally I play some Windows-games from Steam under cxgames (for ideological reasons I'd much rather pay crossover than Microsoft, and avoid dual-booting). The worst part of the experience? The Steam interface itself. Slow to connect, a bit buggy especially under certain window-managers, and in general not a good experience.
Just commented on this in another thread. For some reason, media seems to have disregarded the common press-ethics of protecting identity until sentenced guilty.
I guess in the case of Assange, the temptation to sell some more got too strong.
Or, they thought it's what WikiLeaks would have done.
Last I heard, he is still prosecuted for sexual harassment.
Personally, I think it's a disgrace that newspapers publish names in a case like this. Swedish press tradition around legal cases is innocent until proven guilty, meaning that only after being convicted are people named in press. In this case, however, I guess the promise of selling more papers was irresistable
Or maybe they just think it's what WikiLeaks would have done.
I could easily saturate my 100mbit line, from Giganews or other usenet source, setting up my own news mirror, mirror a few big download sites, or find some other way to waste bandwidth.
My theoretical monthly download capacity would be something like 10MB*3600*24*28 = 24TB, and if that's not enough, there are gigabit upgrades available. However, that's not very interesting, since just the storage cost for 24 TB is much much more than I care to pay.
And, especially, what could I possibly consume that requires those data amounts? Scene-released 720p averages at 7mbit, assuming 1080p averages at 10, and I have to watch 10 simultaneous Full-HD streams around the clock to consume that bandwidth. Who's got the time?
I would take with me what I have in my head. The rest is mostly stuff I could rebuild faster bigger better at the next employer if they need it and I get the chance.
The original poster explicitly asked specifically for means to make non-automated testing more stimulating. Unit-testing and test-driven development is to my knowledge only about automated testing.
Also, one must remember that not all types of software can easily be developed by TTD. I.E. hardware drivers, embedded applications, highly stateful systems, etc. Also I tend to find that TTD (any heavy focus on unittesting, really) can force you to design unnecessarily flexible in order to test components individually, which might also not be apt for highly-performant systems due to overhead.
Where is my mod points? I'd like to mod this up, this is by far a Score:5 Insightful, IMHO the most insightful I've read here - and I almost missed it due to my filter.
Oh, and by the way, sorry for double-posting, but there is a pretty interesting talk on TED recently on the subject of copyright in the clothing industry. The point being that there is none, and how the clothing industry have managed without it.
So, to avoid liability, or investing heavily into a product that will get a cease-and-desist before returning on investment, you better get your lawyer working before thinking about your little startup company.
Many nice observations there, but instead of ruining it for everybody by trying to rephrase them, just spend 2 minutes and watch the beginning. You'll likely watch the rest too.;)
Sorry. US yearly spending on cosmetics is 41bn EUR. So it's more than 10 times as high as Cancer research.
Also interesting, In 2009, Colipa bragged how they spent 25mn EUR on research for finding alternatives to animal testing. 0.06% of their yearly income.
You've got a point. On the other hand, I'm constantly shocked at things like how US Cancer Research gets ~$4.81bn yearly , while I.E. the cosmetics industry, just on cosmetic products, just in the US, chew ~$41bn yearly.
Except clock-cycles, which is what you get from your equation, is also not a good measurement of "computer usage".
However, Google Tech Talks had a rather nice explanation of the algorithm and core mechanics for solving the problem a couple of years ago. Quite interesting for anyone in supercomputing, or just plain old curiosity.
Grabbing foreign citizens in other countries against local laws and extradition treaties.
Happened here in Sweden. There was quite a notable case where the Swedish government extradited two Egyptians via CIA to Egypt (attemptedly in secret, but it was discovered), countrary to local and international law.
FreeNet, is basically designed for this; secured, authenticated yet anonymous publishing and communication. Designed for when your opinion may even get you killed.
Anecdote; I suck at teaching and presentation but occasionally, I have to do it anyways.
Once I had to do a 40minute presentation, prepared rigorously with material research and powerpoint slides. At the time of presentation however, there was a mixup, and I got the wrong presentation-file with me, so I had to draw the important sketches during the presentation, and talk from memory.
According to the feedback both at the time, and followup months after, it was one of my most appreciated presentations ever.
However, in my workplace, we have a pretty interesting setup, where corporate IT operate their Windows-based platform, and pretty strictly try to adhere to that.
In the local technology department, where I work, we however have end-user production on a mostly-streamlined CentOS platform. Whenever we encounter related needs for supporting systems (Customer, support, monitoring, issue-tracking,...) we often take it on ourselves, instead of pushing IT, and then look for solution that fit our hosting platform, to not disperse operations staff too much.
So within the company, we mainly have two fairly streamlined platforms, with two separate staffings with skills for the respective platform. It works out quite well most of the time. Occasionally we readjust the responsibilities, when support-systems or business logic shifts focus/target group, sometimes leading to a systems-change, but only a system at a time.
Yes, and it's the job of IT to do an long-term impact analysis, and explain it to procurement. If the estimated long-term impact (for instance, staying on older, less productive and less safe browsers for longer time, requiring extra support staff, etc.), is still outweighed by the positive effect of using the "odd ball" software, then the software should obviously be used. The point is that the "odd-ball-software upgrade" should be budgeted a couple of years away, and "no, it's too expensive to upgrade related software" should not be a valid argument, since the cost were already budgeted in the acquisition of the blocking software.
Actually, most likely IT will just refuse to support it, so then the area buying the software will just hire some consultant to come in one-time and deploy a small server farm in somebody's closet.
Agreed. I've been there, on both sides, and I've been guilty of plain refusal, but it's hardly ever a good use of anyone:s time.
Plain refusal to cooperate means you just gain a new enemy, and it will almost always bite you or your organisation in the behind. Instead your response should offer motivated options; "Doing this is feasible, but will have these consequences:" If the organisation accepts those consequences (support costs, limited agility,...) great. If not, then go back to the vendor and see if they can offer a solution. (Fix the problem, economic compensation,...)
My point is that in my experience, these long-term effects are too often neglected, and "it works now" is usually good enough. For a small company with lower systems-complexity, that may be OK, but for larger organisations, it will cause problems.
Slashdot Mirror
Except more complex software is more error-prone, especially in the unpredictable failure-conditions you want to catch.
You really do not want to be vulnerable to lighting by having an antenna, and you probably want a highly error-tolerant processing architecture, to cope with the higher-levels of radiation. That may compromise performance and programming methodology. (I.E. read about single-bit-errors in normal PC-architectures)
Keep it simple, and make sure to add a tracking-beacon so they find the darn thing.
I can see it in front of me now;
Salesperson: Look, our state-of-the art FDR records directly to the cloud! No data-box to find and retrieve, no fuss, and MapReduce there immediately after the accident to start the analysis.
Airline manufacturer: Sounds great, I'll buy.
A couple of months pass, the first aircraft goes down...
Air Crash Investigator: It seems we had no 3g connection minutes before the crash, and the plane was pointing away from the satellite. We haven't got a clue.
Salesperson: Oh well, better luck next time.
There's an argument for rugged well-tested simple designs here.
The "grub should'nt do this" point comes up a few times here.
However, one thing that eludes me is; is there really any sane area to write bootloader-code above MBR-sizes in a PC-BIOS boot system?
What happens if you gpt-partition your disk? Does the windows programs disregard and overwrite in the same way?
Or is the only decent way out of this EFI?
This doesn't seem to cover license-claims for implementations of H264, which most of the debate has been about lately? What happens to that, has it too been extended?
Every piece of specialized software will fail occasionally. Not talking highly productized software with a gazillion beta-testers, but the ERP integration layers and other type of software that has been more or less written for a single, or for a few customers.
No developer, and no amount of synthetical testing can ever cover all the possible angles in these highly complex systems the way the real world can. This is especially noticable for integration systems, highly dependent on external environment.
Some of this software is highly mission-critical, when it stops, business stops. In these cases, dev-access is probably the sound way to troubleshoot, and get things running smoothly again.
"The amount of electron-spitting components doubles in density every 18 months."
There you go. I call it Ulriks law. Spread the word.
There are a few gaming companies producing Linux-versions, and selling on Steam. Aquaria, Osmos and World of Goo I know for sure, but I think also the Penumbra series, Savage, Lugaru, Overgrowth, Bridge Builder, Dark Horizons, Machinarium, Serious Sam, and probably some more are available on Steam (and certainly on Linux).
Especially, I think partnership with Cedega or CrossOver could have been one way to go about it.
Personally I play some Windows-games from Steam under cxgames (for ideological reasons I'd much rather pay crossover than Microsoft, and avoid dual-booting). The worst part of the experience? The Steam interface itself. Slow to connect, a bit buggy especially under certain window-managers, and in general not a good experience.
The second part often disturbs me in Swedish press as well. Often the wording in the press clearly states guilty, well before the verdict is declared.
Just commented on this in another thread. For some reason, media seems to have disregarded the common press-ethics of protecting identity until sentenced guilty.
I guess in the case of Assange, the temptation to sell some more got too strong.
Or, they thought it's what WikiLeaks would have done.
Last I heard, he is still prosecuted for sexual harassment.
Personally, I think it's a disgrace that newspapers publish names in a case like this. Swedish press tradition around legal cases is innocent until proven guilty, meaning that only after being convicted are people named in press. In this case, however, I guess the promise of selling more papers was irresistable
Or maybe they just think it's what WikiLeaks would have done.
I could easily saturate my 100mbit line, from Giganews or other usenet source, setting up my own news mirror, mirror a few big download sites, or find some other way to waste bandwidth.
My theoretical monthly download capacity would be something like 10MB*3600*24*28 = 24TB, and if that's not enough, there are gigabit upgrades available. However, that's not very interesting, since just the storage cost for 24 TB is much much more than I care to pay.
And, especially, what could I possibly consume that requires those data amounts? Scene-released 720p averages at 7mbit, assuming 1080p averages at 10, and I have to watch 10 simultaneous Full-HD streams around the clock to consume that bandwidth. Who's got the time?
I would take with me what I have in my head. The rest is mostly stuff I could rebuild faster bigger better at the next employer if they need it and I get the chance.
The original poster explicitly asked specifically for means to make non-automated testing more stimulating. Unit-testing and test-driven development is to my knowledge only about automated testing.
Also, one must remember that not all types of software can easily be developed by TTD. I.E. hardware drivers, embedded applications, highly stateful systems, etc. Also I tend to find that TTD (any heavy focus on unittesting, really) can force you to design unnecessarily flexible in order to test components individually, which might also not be apt for highly-performant systems due to overhead.
Where is my mod points? I'd like to mod this up, this is by far a Score:5 Insightful, IMHO the most insightful I've read here - and I almost missed it due to my filter.
Oh, and by the way, sorry for double-posting, but there is a pretty interesting talk on TED recently on the subject of copyright in the clothing industry. The point being that there is none, and how the clothing industry have managed without it.
http://www.ted.com/talks/johanna_blakley_lessons_from_fashion_s_free_culture.html
suppose you own a small business that has created a remarkable new widget.
Also, assuming you managed to sidestep any existing patents already covering creating widgets;
http://www.google.com/patents?q=widget&btnG=Search+Patents
Notable examples are; ...
Widget Styling and Customization - http://www.google.com/patents/about?id=DpW_AAAAEBAJ&dq=widget
Widget Databases - http://www.google.com/patents/about?id=aLGgAAAAEBAJ&dq=widget
Self-adapting Widgets - http://www.google.com/patents/about?id=3O-aAAAAEBAJ&dq=widget
Synchronizing Widgets - http://www.google.com/patents/about?id=rXKiAAAAEBAJ&dq=widget
Sports-related Widgets - http://www.google.com/patents/about?id=nw6CAAAAEBAJ&dq=widget
Interactive Video Widgets - http://www.google.com/patents/about?id=sAqqAAAAEBAJ&dq=widget
It would definitely take some time going through all the potential patents, see which may still hold vailidity and which have prior art. Also, the above search of course only covers the patents that includes the keyword widget. There may be many more conflicting patents under other categories. Such as, if you got the idea to include a radial menu, covered by http://www.google.com/patents/about?id=lobNAAAAEBAJ&dq=radial+menu and http://www.google.com/patents/about?id=rgHPAAAAEBAJ&dq=radial+menu.
So, to avoid liability, or investing heavily into a product that will get a cease-and-desist before returning on investment, you better get your lawyer working before thinking about your little startup company.
Rory Sutherland touched this subject earlier of Ted. The 12-minute talk is here; http://www.ted.com/talks/rory_sutherland_sweat_the_small_stuff.html
Many nice observations there, but instead of ruining it for everybody by trying to rephrase them, just spend 2 minutes and watch the beginning. You'll likely watch the rest too. ;)
Sorry. US yearly spending on cosmetics is 41bn EUR. So it's more than 10 times as high as Cancer research.
Also interesting, In 2009, Colipa bragged how they spent 25mn EUR on research for finding alternatives to animal testing. 0.06% of their yearly income.
You've got a point. On the other hand, I'm constantly shocked at things like how US Cancer Research gets ~$4.81bn yearly , while I.E. the cosmetics industry, just on cosmetic products, just in the US, chew ~$41bn yearly.
Except clock-cycles, which is what you get from your equation, is also not a good measurement of "computer usage".
However, Google Tech Talks had a rather nice explanation of the algorithm and core mechanics for solving the problem a couple of years ago. Quite interesting for anyone in supercomputing, or just plain old curiosity.
http://www.youtube.com/watch?v=WQw7c-PliB4
Grabbing foreign citizens in other countries against local laws and extradition treaties.
Happened here in Sweden. There was quite a notable case where the Swedish government extradited two Egyptians via CIA to Egypt (attemptedly in secret, but it was discovered), countrary to local and international law.
FreeNet, is basically designed for this; secured, authenticated yet anonymous publishing and communication. Designed for when your opinion may even get you killed.
Anecdote; I suck at teaching and presentation but occasionally, I have to do it anyways.
Once I had to do a 40minute presentation, prepared rigorously with material research and powerpoint slides. At the time of presentation however, there was a mixup, and I got the wrong presentation-file with me, so I had to draw the important sketches during the presentation, and talk from memory.
According to the feedback both at the time, and followup months after, it was one of my most appreciated presentations ever.
Not really no.
However, in my workplace, we have a pretty interesting setup, where corporate IT operate their Windows-based platform, and pretty strictly try to adhere to that.
In the local technology department, where I work, we however have end-user production on a mostly-streamlined CentOS platform. Whenever we encounter related needs for supporting systems (Customer, support, monitoring, issue-tracking, ...) we often take it on ourselves, instead of pushing IT, and then look for solution that fit our hosting platform, to not disperse operations staff too much.
So within the company, we mainly have two fairly streamlined platforms, with two separate staffings with skills for the respective platform. It works out quite well most of the time. Occasionally we readjust the responsibilities, when support-systems or business logic shifts focus/target group, sometimes leading to a systems-change, but only a system at a time.
...then IT will be asked to support it.
Yes, and it's the job of IT to do an long-term impact analysis, and explain it to procurement. If the estimated long-term impact (for instance, staying on older, less productive and less safe browsers for longer time, requiring extra support staff, etc.), is still outweighed by the positive effect of using the "odd ball" software, then the software should obviously be used. The point is that the "odd-ball-software upgrade" should be budgeted a couple of years away, and "no, it's too expensive to upgrade related software" should not be a valid argument, since the cost were already budgeted in the acquisition of the blocking software.
Actually, most likely IT will just refuse to support it, so then the area buying the software will just hire some consultant to come in one-time and deploy a small server farm in somebody's closet.
Agreed. I've been there, on both sides, and I've been guilty of plain refusal, but it's hardly ever a good use of anyone:s time.
Plain refusal to cooperate means you just gain a new enemy, and it will almost always bite you or your organisation in the behind. Instead your response should offer motivated options; "Doing this is feasible, but will have these consequences:" If the organisation accepts those consequences (support costs, limited agility, ...) great. If not, then go back to the vendor and see if they can offer a solution. (Fix the problem, economic compensation, ...)
My point is that in my experience, these long-term effects are too often neglected, and "it works now" is usually good enough. For a small company with lower systems-complexity, that may be OK, but for larger organisations, it will cause problems.