There is no need for two different file managers - one like evolution or kfm and a completely separate 'file picker' used just for Open and Save dialogue boxes. As others have pointed out, they are a prime example of cruft in user interfaces, there because of the limitations of older systems that couldn't manage a multi-tasking desktop.
Personally I think that RISC OS did this the right way. There is a single file manager program ('Filer') which displays windows showing the content of directories. To load a file, double-click on it or drag it onto an application's icon. To save a file, you drag it from the application to the Filer window. This is much more intuitive and also quicker - for example you can have two Filer windows for two directories and load or save files in one or the other without clicking about in the file selector to go up one directory and into another.
This style of drag and drop saving has been implemented in the ROX desktop, but so far the other desktop environments haven't picked up on it, preferring to imitate Windows.
WTF? You're saying it is a bad thing for them to try to make better distributions? That Xandros should deliberately make their Linux version worse to avoid taking sales from Red Hat?
For home users who need Office, internet, and a gaming platform, Windows 98SE upgraded and patched does everything they need.
Exactly. And it won't be possible to keep it patched now that Microsoft has stopped supporting it. It may be hard to upgrade a Win98SE system to existing levels too, if stuff disappears from the MS website.
Huh? How do you 'clone files' except by 'making exact copies down to the bits'? If the filesystem has lots of free space then it will be quicker to copy just the files, but then you're not getting an exact copy of the filesystem (including any weird licence managers that lurk in strange places).
dump + gzip? Is there a 'dump' command for Windows?
I'd say an OS that can't run third-party applications without gunging up the machine is faulty OS. It's the operating system's job to isolate applications from the effects of each other. Bad drivers, okay, that's a good reason for a box to become flaky; but there is no reason at all why installing a new instant messaging program or video codec should screw up completely unrelated things.
I don't really see the point in an article aimed towards lawyers. Lawyers with any training in copyright law are unlikely to misunderstand the various free software licences unless they are paid to do so.
In any case the article has a major mistake:
When we speak of Free Software, we are not talking about freeware, i.e., software that is essentially in the public domain. Rather, we are talking about software that is licensed under the precepts of the Free Software Foundation ("FSF") and its flagship GNU General Public License.
This is a strange definition of several terms. Firstly, the FSF itself treats free software as a much wider range than GPL. The LGPL, BSD, X11, Qt, Netscape and other licences are all considered free software licences. I thought it was only Slashdot posters who misrepresented the FSF as standing for GPL-only.
Secondly, software that is in the public domain (and remember this is aimed at lawyers, so take the legal meaning of public domain, for example the works of Shakespeare) is certainly free software if it has source code. You can do anything with the code including relicensing it under something like the GPL. But this is not the normal definition of 'freeware', which is usually taken to mean binary-only software that is copyrighted and may have a licence forbidding resale for profit.
The other Groklaw article on 'GPL myths' was much better, and shorter.
Yes, and I've had NT 3.51 systems running for years (not continuously, but regular use) with no problems at all and no crashes whatsoever. There are always anecdotes about 'good' Windows installations, but you have to take the big picture. Ask anyone who spends their time looking after Windows user desktops (and hasn't locked them down to the point of making it impossible to run any new applications or do any programming). And these people have not done anything especially odd or stupid to make their machines unreliable. It just goes wrong mysteriously by itself, or at least the default settings and philosophy make it seem so.
In my experience, even Win2k does deteriorate and need reinstalling - after the reinstall the machine is much snappier but then it starts to slow down again. I was mostly thinking of 95/98/ME though. Since Internet Explorer comes with the OS it's fair to count its failings as failings of Windows, especially since we count any security holes of exim or sshd or bind as weaknesses in Linux.
But once you've installed Linux or FreeBSD and set up whatever obscure drivers you need, it normally stays running. It doesn't get slower and slower until it needs reinstalling, or require you to reinstall things because they have 'expired', or get infected with malware (unless you deliberately do something stupid). You have to keep up to date with security fixes, but with a good distribution that is mostly automatable and security fixes won't break random other things. When you do need to reinstall, most things are included on the CD so you don't need to manually step through a dozen slightly different installation wizards to get your applications back.
Besides, if your time is worth more than $5/hour you can simply buy supported, good-quality hardware, install Linux or BSD on it, and still make a net saving compared to Windows. (Assuming that this blogger's total of 11 hours per month is representative, and that a non-Windows system would avoid most of it, which is true.)
For every one 'techno-competent' Slashdot reader who attacks the spammer, there will be ten who get fooled by a Joe job and attack some innocent party.
I know about ISO 8601, my point is that deciding to use an ISO 8601 string for all internal datetime representation is no simpler than deciding to use seconds since 1970. And as you mention, there are variations within the standard (space versus 'T' for separation) and the added complication of timezones which are not needed for a basic internal format (just store everything in UTC internally).
If you store it as a text string, you still have to 'worry about how it is stored'. There are a hundred different date and time string formats. OK, you could pick one and use that everywhere, but how would that be easier than picking a single numeric representation such as seconds since 1970?
Honestly, it's hard to believe that it took PHD "rocket scientists" to come to the conclusion that email is probably better interfaced as a forum. We've all known that for years.
Indeed: see gmane.org. (Which itself has a strong cultural inheritance from Gnus, the Emacs newsreader that lets you read mail and other things as though they were newsgroups.)
In the story, the supposed faults of electronic mail - that it is easy to view one message but hard to set it in context, to see who is saying what to whom - could be fixed or at least greatly reduced by proper quoting and attribution. If Microsoft changed Outlook and Outlook Express to make it easy and encouraged for users to do concise, non-Jeopardy quoting followed by a reply, rather than typing a few words then followed by masses of irrelevant autoappended gunk, they'd do the net a big favour.
(See OE-QuoteFix - but I don't know of an equivalent for Outlook.)
Feh. The only proper way to read them is to parallelize in chunks of three books at a time, reading every alternate book in reverse chapter order and then randomizing the paragraphs in every fourth book.
Notice what Trolltech has done with their QT widgets with the signal/slot pre-compilation process (they had to work around a limitation of the language),
Sigslot is a pure C++ implementation of the same idea, suggesting that the language isn't so deficient you _have_ to use a preprocessor (although it may have been back in the 1990s when Qt was developed).
If you look at Hubbards' website you'll see they have a whole range of crazy flavoured cereals (including an orange flavoured one which is not (quite) as disgusting as it sounds). But in this case couldn't you sprinkle some instant coffee granules on your cornflakes / muesli before adding milk?
I think most of the gotchas are related to data integrity and things like declaring a foreign key constraint which is then ignored under some circumstances, or columns which get assigned values you didn't explicitly ask for. A particular app may run fine, but when you ask the database to enforce something like a foreign key constraint, the value of that is that the database makes sure it will always hold no matter what. If you can't trust the system not to do random things with your data (and the gotchas list seems to suggest you cannot), then the foreign key checking (part of ANSI SQL) becomes almost worthless.
(FWIW, I do 'insert into... select x, y from...' all the time. I'm sure your app is fine, but that's because it is using only a subset of standard SQL. You can't pick a particular example, say 'it works for me', and use that to conclude the problems are not serious. If that were true we'd have to class MS VC++ 6 as a standards-compliant C++ compiler or Netscape 3.0 as a web browser that supports CSS.)
Sure, IO performance may suck, but that's not a reason to support only a subset of ANSI SQL. Not if you already have a database engine that does the job properly, and reduced speed is the only difference moving it from a server to a handheld. For small databases RAM should be big enough to cache most things anyway, so I/O speed doesn't matter that much. You could probably put the transaction log in RAM too.
Slashdot Mirror
There is no need for two different file managers - one like evolution or kfm and a completely separate 'file picker' used just for Open and Save dialogue boxes. As others have pointed out, they are a prime example of cruft in user interfaces, there because of the limitations of older systems that couldn't manage a multi-tasking desktop.
Personally I think that RISC OS did this the right way. There is a single file manager program ('Filer') which displays windows showing the content of directories. To load a file, double-click on it or drag it onto an application's icon. To save a file, you drag it from the application to the Filer window. This is much more intuitive and also quicker - for example you can have two Filer windows for two directories and load or save files in one or the other without clicking about in the file selector to go up one directory and into another.
This style of drag and drop saving has been implemented in the ROX desktop, but so far the other desktop environments haven't picked up on it, preferring to imitate Windows.
WTF? You're saying it is a bad thing for them to try to make better distributions? That Xandros should deliberately make their Linux version worse to avoid taking sales from Red Hat?
Huh? How do you 'clone files' except by 'making exact copies down to the bits'? If the filesystem has lots of free space then it will be quicker to copy just the files, but then you're not getting an exact copy of the filesystem (including any weird licence managers that lurk in strange places).
dump + gzip? Is there a 'dump' command for Windows?
I'd say an OS that can't run third-party applications without gunging up the machine is faulty OS. It's the operating system's job to isolate applications from the effects of each other. Bad drivers, okay, that's a good reason for a box to become flaky; but there is no reason at all why installing a new instant messaging program or video codec should screw up completely unrelated things.
In any case the article has a major mistake:
This is a strange definition of several terms. Firstly, the FSF itself treats free software as a much wider range than GPL. The LGPL, BSD, X11, Qt, Netscape and other licences are all considered free software licences. I thought it was only Slashdot posters who misrepresented the FSF as standing for GPL-only.
Secondly, software that is in the public domain (and remember this is aimed at lawyers, so take the legal meaning of public domain, for example the works of Shakespeare) is certainly free software if it has source code. You can do anything with the code including relicensing it under something like the GPL. But this is not the normal definition of 'freeware', which is usually taken to mean binary-only software that is copyrighted and may have a licence forbidding resale for profit.
The other Groklaw article on 'GPL myths' was much better, and shorter.
Yes, and I've had NT 3.51 systems running for years (not continuously, but regular use) with no problems at all and no crashes whatsoever. There are always anecdotes about 'good' Windows installations, but you have to take the big picture. Ask anyone who spends their time looking after Windows user desktops (and hasn't locked them down to the point of making it impossible to run any new applications or do any programming). And these people have not done anything especially odd or stupid to make their machines unreliable. It just goes wrong mysteriously by itself, or at least the default settings and philosophy make it seem so.
In my experience, even Win2k does deteriorate and need reinstalling - after the reinstall the machine is much snappier but then it starts to slow down again. I was mostly thinking of 95/98/ME though. Since Internet Explorer comes with the OS it's fair to count its failings as failings of Windows, especially since we count any security holes of exim or sshd or bind as weaknesses in Linux.
But once you've installed Linux or FreeBSD and set up whatever obscure drivers you need, it normally stays running. It doesn't get slower and slower until it needs reinstalling, or require you to reinstall things because they have 'expired', or get infected with malware (unless you deliberately do something stupid). You have to keep up to date with security fixes, but with a good distribution that is mostly automatable and security fixes won't break random other things. When you do need to reinstall, most things are included on the CD so you don't need to manually step through a dozen slightly different installation wizards to get your applications back.
Besides, if your time is worth more than $5/hour you can simply buy supported, good-quality hardware, install Linux or BSD on it, and still make a net saving compared to Windows. (Assuming that this blogger's total of 11 hours per month is representative, and that a non-Windows system would avoid most of it, which is true.)
For every one 'techno-competent' Slashdot reader who attacks the spammer, there will be ten who get fooled by a Joe job and attack some innocent party.
Ah, I see that cpd is included in the sources for pmd, that's what confused me.
I see the link to a page about CPD, but I couldn't find a way to download that tool... any ideas?
I know about ISO 8601, my point is that deciding to use an ISO 8601 string for all internal datetime representation is no simpler than deciding to use seconds since 1970. And as you mention, there are variations within the standard (space versus 'T' for separation) and the added complication of timezones which are not needed for a basic internal format (just store everything in UTC internally).
If you store it as a text string, you still have to 'worry about how it is stored'. There are a hundred different date and time string formats. OK, you could pick one and use that everywhere, but how would that be easier than picking a single numeric representation such as seconds since 1970?
But in speech, most people nowadays say 'twenty-four' not 'four and twenty'...
They've obviously been taking lessons from SCO on how to extract 'facts' from source code.
Indeed: see gmane.org. (Which itself has a strong cultural inheritance from Gnus, the Emacs newsreader that lets you read mail and other things as though they were newsgroups.)
In the story, the supposed faults of electronic mail - that it is easy to view one message but hard to set it in context, to see who is saying what to whom - could be fixed or at least greatly reduced by proper quoting and attribution. If Microsoft changed Outlook and Outlook Express to make it easy and encouraged for users to do concise, non-Jeopardy quoting followed by a reply, rather than typing a few words then followed by masses of irrelevant autoappended gunk, they'd do the net a big favour.
(See OE-QuoteFix - but I don't know of an equivalent for Outlook.)
Scientific progress goes BOINC?
(according to Google this joke is not original, but what the hell)
Feh. The only proper way to read them is to parallelize in chunks of three books at a time, reading every alternate book in reverse chapter order and then randomizing the paragraphs in every fourth book.
Sigslot is a pure C++ implementation of the same idea, suggesting that the language isn't so deficient you _have_ to use a preprocessor (although it may have been back in the 1990s when Qt was developed).
If you look at Hubbards' website you'll see they have a whole range of crazy flavoured cereals (including an orange flavoured one which is not (quite) as disgusting as it sounds). But in this case couldn't you sprinkle some instant coffee granules on your cornflakes / muesli before adding milk?
I think most of the gotchas are related to data integrity and things like declaring a foreign key constraint which is then ignored under some circumstances, or columns which get assigned values you didn't explicitly ask for. A particular app may run fine, but when you ask the database to enforce something like a foreign key constraint, the value of that is that the database makes sure it will always hold no matter what. If you can't trust the system not to do random things with your data (and the gotchas list seems to suggest you cannot), then the foreign key checking (part of ANSI SQL) becomes almost worthless.
... select x, y from ...' all the time. I'm sure your app is fine, but that's because it is using only a subset of standard SQL. You can't pick a particular example, say 'it works for me', and use that to conclude the problems are not serious. If that were true we'd have to class MS VC++ 6 as a standards-compliant C++ compiler or Netscape 3.0 as a web browser that supports CSS.)
(FWIW, I do 'insert into
Sure, IO performance may suck, but that's not a reason to support only a subset of ANSI SQL. Not if you already have a database engine that does the job properly, and reduced speed is the only difference moving it from a server to a handheld. For small databases RAM should be big enough to cache most things anyway, so I/O speed doesn't matter that much. You could probably put the transaction log in RAM too.