I've been working on a free alternative (see http://www.malsmith.net/yori/). I won't be able to replicate all of bash, but there's plenty of nice things that can be incorporated into a native Windows shell, and it's sure keeping me busy.
Finding a group of people facing a similar problem with skills to contribute.
Demonstrating enough value to them to justify their effort.
These are both really (!) hard things to do. The software market is fairly crowded these days, so demonstrating you can beat existing offerings is hard. And although the Internet allows people to search fairly well, finding people facing similar problems is still really hard. It's not like you can show up on forums of a similar product, pitch your idea for a different thing, and expect to be well received.
Realistically, don't start a software project if you're unwilling to be the only one working on it.
(Personally I'm trying to build a decent native code Windows command prompt, and I can't believe others really think CMD is wonderful, or that managed environments are fun to use given their performance, or that sitting on emulation layers is a great way to interact with an OS. But the set of people who are passionate and skilled enough to act isn't that big, and they don't all hang out in the same place, so I'm still the only contributor, and I don't expect that to change imminently.)
A decade ago, I wrote a socks server and posted it to Sourceforge. It does exactly what it says it will do, and it was so good and convenient that malware authors found it to be a useful payload to drop on machines to get a backdoor into them. So then virus scanners flagged it as malware, and sourceforge trusts those, and then they deleted the current version of the binary. Now that page has big scary warnings about software that plainly does what it says with all the source there to prove it (see it for yourself - https://sourceforge.net/projec... ).
I know these guys are trying to win back trust, but trust is hard. Trusting heuristic based scanners is optimistic. Making allegations about software and its authors on the basis of a heuristic can be downright offensive.
(Along similar lines, chocolatey is now flagging my directory enumerator because one out of 57 virus scanners heuristically thought crawling a disk is suspicious - https://www.virustotal.com/en/....)
Notwithstanding its clear bugs and usability flaws, Windows 3.0 was the first version to allow developers to fully use the PC's memory and processor. It was the first version I developed for, and I continued to target it for many years after users had moved on to the more stable, polished and complete 3.1 and 3.11 versions. Although, it was a pain to exit to DOS, compile something, launch Windows, execute it, and repeat.
And developers using file access APIs in Windows can get case sensitive behavior. It's just a single FILE_FLAG_POSIX_SEMANTICS flag to CreateFile and friends. It's actually easier to do that in Windows because it doesn't require a reformat and reinstall.
For the record, note that this is no longer true as of Windows XP and later. FILE_FLAG_POSIX_SEMANTICS is overridden by the ObCaseInsensitive registry key, which defaults to 1, meaning that all opens will be case insensitive. See this kb article for example. Changing this requires a reboot, but not a reformat.
I have a prerelease build of OS X built with gcc 2.7. I remember the Apple engineers wanted to ship the original OS X with gcc 3.x, but 'it just wouldn't make it' and they used 2.96 instead. OS X 10.0 and 10.1 were built on that compiler.
I smell email address harvester. As has already been pointed out, Linux people and Microsoft people know the answers to these questions. That last question is the only one with real commercial value.
Is Forbes genuinely incapable of understanding that the whole point of Open Source is that it represents a parallel software development strategy that is opposed to the conventional business paradigm of proprietary I.P., or are they engaged in conscious propaganda in defense of the status quo?
Or, are Forbes acting in the mistaken belief that the only thing that matters to Open Source developers is popularity, and they are telling us how to be more popular among big business? I think they just don't get the "principle" of Free Software at all.
Or even direct SSL? I'm sure there'd be a way of incorporating GPG signing into an HTTP header, but I can't think off the top of my head of a secure way of doing it.
If your company is modifying web documents, even if they have a right to do it, surely you have a right to know they're doing it and decide what weight to give to the information.
No one can tolerate DDoS attacks and other kinds of attacks in this Information Age economy that relies so heavily on the Internet.
Although it takes a different form, that's what SCO is trying to do in convincing people that they must license SCO software to use Linux. I don't support DDoS as a concept, including when SCO does it.
At a minimum, IP sources should be checked to assure that copyright contributors have the authority to transfer copyrights in the code contributed to Open Source.
So SCO ensures that all employees have the authority to transfer copyrights to code contributed in their private developments, does it?
This is what global corporations will require. And it is these customers who will determine the ultimate fate of Open Source - not SCO, not IBM, and not Open Source leaders.
I decide what gets into MyProjects, not global corporations. Hell, I don't even know who uses my products. That's the point with open source.
In copyright law, ownership cannot be transferred without express, written authority of a copyright holder.
Look, I'm a Law student. Anybody who's read the GPL knows that it doesn't transfer ownership at all. Ownership is retained; certain rights are licensed in respect of that ownership, including the right to freely redistribute on the same terms. This statement is simply not legally correct.
It is easier for some in the Open Source community to fire off a "rant" than to sit across a negotiation table.... respect for intellectual property is not optional - it is mandatory.
Listen to yourself! Declaring something to be "mandatory" is not negotiation! While what's said here is technically true, the inference of "respect" is that that respect has not been honoured, a factually dicey proposition.
As the owner of a small software business, I feel comfortable with the fact that whilst I cannot sue Microsoft's ass if something goes terribly wrong; neither can my customers sue my ass.
Maybe they can't sue you for any implied warranty, but they can sue you in Tort (negligence.) This non-litigious culture that has developed around software is analagous to the eye of a storm: You don't see the litigation, but the potential is there brewing, and it may break at short notice. If it does, the consequences for the industry are immense.
I agree with your comments re lenience: but lenience is not absolute. Negligence is still negligent.
Further, it's there to prevent access by a licensee that the licensee may be entitled to.
In Australia, for example, we have a limited set of rights in relation to Computer Software in the Copyright Act 1968 (Cth); preventing a licensee from exercising rights by encryption is, in effect, trying to subvert the operation of the law; as much as DRM companies would like to think so, they do not have a monopoly in determining how their products can/shall/will be used.
I agree - TT's scheme makes a lot of sense - but (obviously) I can't pay $1500 for something without commercialising it. That means the cool features of Qt (Win32/Mac compatibility) are pretty much off-limits to Open Source developers (last time I checked, there is a non-commercial version of Qt for Win32 for download, but its licencing is GPL-incompatible.)
GTK+ leaves open the Windows option, and the closed-source option (note: only for a new version. Once GPLed, always GPLed.)
No; as long as each stream continues to back off, the emergent properties of such a throttling regime will allow for an equilibrium to be reached.
...Eventually. If in Fast TCP, instead of halving the send speed, you reduce it by, say, 2% - if the number of packets that have been sent to cause an overflow are excessive, it is unlikely that a small reduction will prevent packets being discarded, because they are still being pumped in faster than a router can buffer them.
So if the saturated router can clear enough buffer space after some time, the Fast TCP implementation may have to resend the packet five times - whereas the traditional TCP only has to send it again once. This is because Fast TCP will increase the number of discards. Now, it may still perform faster (to the end user) because it won't have to go so far to regain its previous send rate. But if it increases the number of discards in the network as a whole efficiency is decreased.
Yes, I can understand that - but the reason the LGPL exists is for this exact reason. If every binary compiled with gcc had to be GPLed, then the FSF would be happy - but only if that meant the binaries were opened. It would be far more likely that proprietary software would not be developed on GPL OSes instead.
While I'm no fan of proprietary software, it does serve as a benchmark to compete with. Netscape 4, for instance, allowed Linux users to have a graphical browser until free/open browsers became available. VMware is now being replaced (by Bochs) and so on.
So IMO, the LGPL should be used on 'core' libraries that provide essential functionality - to preserve the ability of proprietary vendors to create applications on free OSes. The GPL should be used on complex, extension libraries that provide added functionality, because it would not be desirable that sophisticated proprietary binaries could be developed without substantial and significant investments of proprietary source.
I'm not so sure - I prefer to develop in GTK+ because I know I have the flexibility to use my code however I like in the future. My code is currently all GPLed, and I have every intention of maintaining it GPLed, but if there was a good reason down the track to make a closed fork, GTK+ gives me that option.
Yes, there's much more work to be done in OpenOffice. I prefer the smaller, faster, cleaner alternatives (see KOffice), although these tend to have much worse file format support. Forcing some of these specifications open would be beneficial.
...but back to the point. They would be beneficial not just to KOffice, or AbiWord, or any other tool; having tools with increased compatibility increases consumer choice and competition, not just for free software, but commercial software as well. This bill seems to me to be quite logical, in the interests of keeping government data readable, and in the interests of promoting market diversity.
More to the point, the reason you lost a packet is likely to be network-wide congestion. If one connection doesn't back off its sending, but keeps sending at the same rate, it relies on other connections slowing down to allow it to do so.
If no connection slowed down, packets would be being dropped all over the place, and total network throughput would decrease.
In other words, this Fast TCP idea seems inherently selfish and will only improve performance for one user over others, not for all users if adopted.
But X is becoming more bandwidth-hungry as developers plug in more complex, animated, client-side widget toolkits. I think I've said before that widget toolkits are better off on the X display; but nobody listens to me:)...
Tweaking the settings of your widget toolkits should yield far superior bandwidth results than the defaults in most cases.
Now all it CW need is some cool technology to take advantage of X in a way that was never possible with Terminal Server...loadbalancing, very little configuration, better scaling, more portability. Yum.
I bought a copy of the original 'The Microsoft Office' the other day for a joke. I'd have to check, but I'm quite sure the license says that a seperate license is required for each concurrent 'use' of the software, and a dedicated server doesn't count as a 'use.'
I conceed that there is royalty-free distribution. But there is also a right to sell. Royalty-free redistribution can only be invoked if you can get a copy, and the author can refuse to make it freely available. That doesn't stop end users from replicating it; but it may make software less accessible.
For example, if RedHat 'derives' original authors to form a Linux distribution, it can 'sell' that distribution (sell downloads) which doesn't prevent copying of the distribution, but does restrict its availablility (to some extent.)
Slashdot Mirror
I've been working on a free alternative (see http://www.malsmith.net/yori/). I won't be able to replicate all of bash, but there's plenty of nice things that can be incorporated into a native Windows shell, and it's sure keeping me busy.
These are both really (!) hard things to do. The software market is fairly crowded these days, so demonstrating you can beat existing offerings is hard. And although the Internet allows people to search fairly well, finding people facing similar problems is still really hard. It's not like you can show up on forums of a similar product, pitch your idea for a different thing, and expect to be well received.
Realistically, don't start a software project if you're unwilling to be the only one working on it.
(Personally I'm trying to build a decent native code Windows command prompt, and I can't believe others really think CMD is wonderful, or that managed environments are fun to use given their performance, or that sitting on emulation layers is a great way to interact with an OS. But the set of people who are passionate and skilled enough to act isn't that big, and they don't all hang out in the same place, so I'm still the only contributor, and I don't expect that to change imminently.)
https://www.youtube.com/watch?...
He discusses the pros and cons fairly well.
A decade ago, I wrote a socks server and posted it to Sourceforge. It does exactly what it says it will do, and it was so good and convenient that malware authors found it to be a useful payload to drop on machines to get a backdoor into them. So then virus scanners flagged it as malware, and sourceforge trusts those, and then they deleted the current version of the binary. Now that page has big scary warnings about software that plainly does what it says with all the source there to prove it (see it for yourself - https://sourceforge.net/projec... ).
I know these guys are trying to win back trust, but trust is hard. Trusting heuristic based scanners is optimistic. Making allegations about software and its authors on the basis of a heuristic can be downright offensive.
(Along similar lines, chocolatey is now flagging my directory enumerator because one out of 57 virus scanners heuristically thought crawling a disk is suspicious - https://www.virustotal.com/en/... .)
Notwithstanding its clear bugs and usability flaws, Windows 3.0 was the first version to allow developers to fully use the PC's memory and processor. It was the first version I developed for, and I continued to target it for many years after users had moved on to the more stable, polished and complete 3.1 and 3.11 versions. Although, it was a pain to exit to DOS, compile something, launch Windows, execute it, and repeat.
And developers using file access APIs in Windows can get case sensitive behavior. It's just a single FILE_FLAG_POSIX_SEMANTICS flag to CreateFile and friends. It's actually easier to do that in Windows because it doesn't require a reformat and reinstall.
For the record, note that this is no longer true as of Windows XP and later. FILE_FLAG_POSIX_SEMANTICS is overridden by the ObCaseInsensitive registry key, which defaults to 1, meaning that all opens will be case insensitive. See this kb article for example. Changing this requires a reboot, but not a reformat.
Office 12 will have open, XML formats, by default. We got the message. http://channel9.msdn.com/ShowPost.aspx?PostID=7332 9
I have a prerelease build of OS X built with gcc 2.7. I remember the Apple engineers wanted to ship the original OS X with gcc 3.x, but 'it just wouldn't make it' and they used 2.96 instead. OS X 10.0 and 10.1 were built on that compiler.
I smell email address harvester. As has already been pointed out, Linux people and Microsoft people know the answers to these questions. That last question is the only one with real commercial value.
Or, are Forbes acting in the mistaken belief that the only thing that matters to Open Source developers is popularity, and they are telling us how to be more popular among big business? I think they just don't get the "principle" of Free Software at all.
Or even direct SSL? I'm sure there'd be a way of incorporating GPG signing into an HTTP header, but I can't think off the top of my head of a secure way of doing it.
If your company is modifying web documents, even if they have a right to do it, surely you have a right to know they're doing it and decide what weight to give to the information.
Although it takes a different form, that's what SCO is trying to do in convincing people that they must license SCO software to use Linux. I don't support DDoS as a concept, including when SCO does it.
So SCO ensures that all employees have the authority to transfer copyrights to code contributed in their private developments, does it?
I decide what gets into My Projects, not global corporations. Hell, I don't even know who uses my products. That's the point with open source.
Look, I'm a Law student. Anybody who's read the GPL knows that it doesn't transfer ownership at all. Ownership is retained; certain rights are licensed in respect of that ownership, including the right to freely redistribute on the same terms. This statement is simply not legally correct.
Listen to yourself! Declaring something to be "mandatory" is not negotiation! While what's said here is technically true, the inference of "respect" is that that respect has not been honoured, a factually dicey proposition.
Maybe they can't sue you for any implied warranty, but they can sue you in Tort (negligence.) This non-litigious culture that has developed around software is analagous to the eye of a storm: You don't see the litigation, but the potential is there brewing, and it may break at short notice. If it does, the consequences for the industry are immense.
I agree with your comments re lenience: but lenience is not absolute. Negligence is still negligent.
Further, it's there to prevent access by a licensee that the licensee may be entitled to.
In Australia, for example, we have a limited set of rights in relation to Computer Software in the Copyright Act 1968 (Cth); preventing a licensee from exercising rights by encryption is, in effect, trying to subvert the operation of the law; as much as DRM companies would like to think so, they do not have a monopoly in determining how their products can/shall/will be used.
I agree - TT's scheme makes a lot of sense - but (obviously) I can't pay $1500 for something without commercialising it. That means the cool features of Qt (Win32/Mac compatibility) are pretty much off-limits to Open Source developers (last time I checked, there is a non-commercial version of Qt for Win32 for download, but its licencing is GPL-incompatible.)
GTK+ leaves open the Windows option, and the closed-source option (note: only for a new version. Once GPLed, always GPLed.)
...Eventually. If in Fast TCP, instead of halving the send speed, you reduce it by, say, 2% - if the number of packets that have been sent to cause an overflow are excessive, it is unlikely that a small reduction will prevent packets being discarded, because they are still being pumped in faster than a router can buffer them.
So if the saturated router can clear enough buffer space after some time, the Fast TCP implementation may have to resend the packet five times - whereas the traditional TCP only has to send it again once. This is because Fast TCP will increase the number of discards. Now, it may still perform faster (to the end user) because it won't have to go so far to regain its previous send rate. But if it increases the number of discards in the network as a whole efficiency is decreased.
Yes, I can understand that - but the reason the LGPL exists is for this exact reason. If every binary compiled with gcc had to be GPLed, then the FSF would be happy - but only if that meant the binaries were opened. It would be far more likely that proprietary software would not be developed on GPL OSes instead.
While I'm no fan of proprietary software, it does serve as a benchmark to compete with. Netscape 4, for instance, allowed Linux users to have a graphical browser until free/open browsers became available. VMware is now being replaced (by Bochs) and so on.
So IMO, the LGPL should be used on 'core' libraries that provide essential functionality - to preserve the ability of proprietary vendors to create applications on free OSes. The GPL should be used on complex, extension libraries that provide added functionality, because it would not be desirable that sophisticated proprietary binaries could be developed without substantial and significant investments of proprietary source.
Just my 2c.
I'm not so sure - I prefer to develop in GTK+ because I know I have the flexibility to use my code however I like in the future. My code is currently all GPLed, and I have every intention of maintaining it GPLed, but if there was a good reason down the track to make a closed fork, GTK+ gives me that option.
Result: I use GTK+ for writing Free software.
Yes, there's much more work to be done in OpenOffice. I prefer the smaller, faster, cleaner alternatives (see KOffice), although these tend to have much worse file format support. Forcing some of these specifications open would be beneficial.
...but back to the point. They would be beneficial not just to KOffice, or AbiWord, or any other tool; having tools with increased compatibility increases consumer choice and competition, not just for free software, but commercial software as well. This bill seems to me to be quite logical, in the interests of keeping government data readable, and in the interests of promoting market diversity.
Agree.
More to the point, the reason you lost a packet is likely to be network-wide congestion. If one connection doesn't back off its sending, but keeps sending at the same rate, it relies on other connections slowing down to allow it to do so.
If no connection slowed down, packets would be being dropped all over the place, and total network throughput would decrease.
In other words, this Fast TCP idea seems inherently selfish and will only improve performance for one user over others, not for all users if adopted.
X just doesn't use that much bandwidth.
But X is becoming more bandwidth-hungry as developers plug in more complex, animated, client-side widget toolkits. I think I've said before that widget toolkits are better off on the X display; but nobody listens to me :)...
Tweaking the settings of your widget toolkits should yield far superior bandwidth results than the defaults in most cases.
Then distribute apps properly.
That's how Wine normally works.
Now all it CW need is some cool technology to take advantage of X in a way that was never possible with Terminal Server...loadbalancing, very little configuration, better scaling, more portability. Yum.
I bought a copy of the original 'The Microsoft Office' the other day for a joke. I'd have to check, but I'm quite sure the license says that a seperate license is required for each concurrent 'use' of the software, and a dedicated server doesn't count as a 'use.'
In other words, old software won't help you.
I conceed that there is royalty-free distribution. But there is also a right to sell. Royalty-free redistribution can only be invoked if you can get a copy, and the author can refuse to make it freely available. That doesn't stop end users from replicating it; but it may make software less accessible.
For example, if RedHat 'derives' original authors to form a Linux distribution, it can 'sell' that distribution (sell downloads) which doesn't prevent copying of the distribution, but does restrict its availablility (to some extent.)
In any event, it's a pretty academic distinction.
-M
In other words,
This must be about money. There's just no logical reason UCSB could possibly come to this conclusion...