> I like how it "started out anonymous" > (read: Hong Kong was part of the UK) and > now is becoming "personally trackable" > (Hong Kong is now part of the PRC).
I tried not to get into the discussion, but this one gets me. No, privacy invasion is not the motivation of personalizing the card. It can be an issue, though. (Yes, I'm a HK citizen.)
Octopus plans to have personalization feature from day 1 of its operation (yes, when UK is in charge of HK), and it is implemented within months after its launch. The motivation for people to personalize their cards is basically for added convenience. E.g., if you lose your card, you can tell the card center about it and make the card a useless plastic. You can link your card with a bank so that recharges are automatically made once the balance gets down to 0. And so on. It is completely optional: if you don't like that, you can instead buy an anonymous card, which is done by pretty much all the people. I doubt any merchant has the restriction that only personalized cards can use their system.
And... Octopus is *not* run by the government here! It is handled by a company, which branch out from MTR (the subway here), to deal with the ever increasing need to shorten time to buy tickets and process the cards before letting passengers to enter their railway systems. There is no linkage between the Octopus companny and the government. If there is any privacy concern, it is due to Octopus, not the government.
Absolute crap. When the software of the whole world is dominated by MS products (at least, many categories of software), and when Taiwan is a major computer equipment user (and also producer), will you remotely think that it has been using only minor $$s in pleasing MS?
And... the "switch" only make sense if you have to pay a lot of money to MS.
What you are talking about is extern, not export. Extern is fully supported by any C++ compiler. Export is not supported by any so far---before this report. Look into The C++ Programming Language (Bjarne Stroustrup) to see what it is. In short, it should allow an uninstantiated template definition to appear in an object file, allowing other object files to somehow use an instantiated version of it without seeing the code again during compilation. Pretty tough requirement indeed.
> I have to find it first, make sure > I actually found an element, and THEN I can > erase it. This means I either keep a local > copy of the iterator, or I call the > find_first algorithm twice.
I still don't know what you are blowing...
#include #include
int main() {
vector test;
for (int i = 0; i 10; ++i)
test.push_back(i);
test.erase(test.begin()+5);
test.erase(find(test.begin(), test.end(), 7));
for (int i = 0; i test.size(); ++i)
cout << test[i] << endl; }
> Just wanted to point out that the STL is > not, in any way, an OO system. It uses > classes, but that's as close as it gets. > There's no use of inheritance, polymorphism > (in the usual OO sense) or any other > "typical OO" features.
Everything in STL is compile-time determined. Then there is no need for using OO concepts like polymorphism. I'd rather count generic programming as a concrete part of OO programming, and say that STL used the alternative way to do OO.
> do they really think that mozilla-m18, which is in debian/stable, is more stable or usable than 1.0?! Or that gnome 1.0.55 is better than gnome-1.4.0.
You get the wrong idea about what stable really is. A stable release is something which is more or less *fixed*, i.e., won't change much. When potato is released, there is no gnome-1.4.0, and gnome-1.2 is just out which is not in production quality yet. So not including them is the right choice.
We will see similar choice made in this stable release. Apache 2 is not stable enough, so 1.3 is shipped. xFree 4.2 is not there long enough yet, so we have XFree 4.1. This is no fault of Debian at all. Stable is for thnigs that "have to be up with no incompatible changes to programs" for a long time. If you don't need such a beast, you can always ignore stable (as I do) and always use testing or unstable.
If you don't use Debian, try to use it. If you are still thinking that everybody should use stable, try the freezed and testing release, or even the unstable release.
Debian might not be the norm in calling something stable only when it mostly won't change. But if you think that you should be using software that are more recent than a year old, then you have no business using Debian stable. It is simply not for that purpose. Use testing or unstable instead.
Is it just me who experience this? On my notebook LCD, I see exactly the same problem described here: at lower resolution, the display is ugly. But on my LCD panel I used on my desktop, it is much less a problem. It seems that the panel does anti-aliasing, but I'm not sure. Of course, not as good as a CRT. But still, much better, to the point that I don't really care.
Just hope that Transmeta will still be living well by the time when OLED display is out there.:)
Re:Well, if anyone had any doubts they are gone no
on
BBC interview with RMS
·
· Score: 1
> Secondary it shows that he isn't interrested in any workable bussinessmodel for free software since he hails the good in people swapping software without paying the creator.
Think again. Remember that for many years RMS pay himself by writing free software. His opposition to business models just go to the extend that "if the business model ruin our freedom, than replace it as soon as we can".
So he call for trashing the current business model. But he doesn't say that all business models are bad.
you must replace your propriety software
with free software, software that lets
you have those freedoms.
Oh... how deceiving... What RMS really says...
To have freedom to live as part of a community, to have the freedom to treat other people decently, you must replace your propriety software with free software, software that lets you have those freedoms.
You can choose whether to have the freedom to live as part of a community. You can choose whether to have the freedom to treat other people decently. But if you choose to do so, you have to replace your proprietry (sic) software. This is exactly correct, at least in the POV of the GNU community. It is not forcing you anything. But it is really good to do so.
With the constraint that the contest should be timed, should be about coding, should be language neutral (to different languages), should be reasonable to be completed within one day, and should be judged objectively rather than subjectively, I see no alternative. Will you care to suggest some?
More importantly, whatever your contest rules, there are some skills needed in programming that is not tested at all. Will you say that it "is not a programming contest" because they don't test the skills in those areas?
Come on. A contest is there to give motivations to work hard, and to gather people so that different bright people can meet each other and share ideas. There is no ``ultimate game rule'' that tests everything. And that is not the objective of the contest.
Penalty: each correctly answer carries a penalty, which is the amount of time elapsed from the beginning of the contest until the correct answer is received, plus 20 minutes times the number of time the same question is submitted with an incorrect answer. So the penalty listed is simply the total penalty (less is better).
Most people who has such a feeling of the contest should go back and check the question again, and think about what will happen if the input is really as large as suggested by the question. You can solve it the Prolog way and use backtracking and so on to find the solution. But the judge will simply answer "time limit exceeded" to your submission of answers, nothing else.
Right. Remember that the problem set is done by 3 contestents, not 1; and all of them share one copy of the questions. It is very frustrating having to read half of the question and wait for your peer to finish reading his part and give you the other half.
> Just as C++ will never replace C (in the > forseeable future), because C++ has its own levels > of indirection and safety which slow it down > (RTTI, virtual tables, etc)
I can't agree with this. RTTI, virtual tables, etc., never shows up until you actually use the feature. If you decide it that way, you can always use the C library subpart of the language and forget about dynamic binding. RTTI is actually rarely used even in the standard library of C++.
On the other hand, if you need the capability, you can't do it better in C. If you write a C language object system, chances are that you will get worse performance, rather than better performance, than a corresponding C++ system, because the developers of the C++ compilers have really put a lot of efforts into supporting the OO concepts in language level.
But your main point is still correct: C++ contains some features that will make it slower and more bulky. Basically, exceptions. If you go into the compiled assembly generated by any recent C++ compiler, you'll find that 20% of the code is generated only to get exception handling correct. While C++ designers will tell you that it won't cause performance problems, in reality it clutter the cache, create a few more branches to fool branch prediction hardware, etc.
But most C++ programs are slow for a completely different reason: the people are more concerned about other things like adding features and maintaining software.
That said, the most serious problem of C++ against C is not performance issue. It is the learning curve issue: C is a small language. You can learn it in a couple of days, and start your venture into each of the library. C++ is a huge language: the language specification goes over 1000 pages, and you'd like to understand the whole standard library before actually working on any real project. That makes it very unattractive for people who don't have their background in C.
There can be only one way for human evolution to really be "over": the natural selection won't work anymore. In other words, everybody survive and can have equal chance to give offsprings, regardness of the shortcoming of the individual. The gene pool will no longer be able to benefit from natural selection.
Unluckily, the current medical advances are really leading us to this sequel.
> Compare with the Linux world, where you have a
> bunch of different distros that many people
> pretend are all the same OS (in spite of the fact
> that file systems are arranged differently, boot
> sequences are different, configuration is
> different, package management is different,
> userland tools are often different, etc.) because
> they happen to use the same kernel.
That makes it extremely easy... just use <add-your-dist-name-here>Linux! I use Debian GNU/Linux, and every installation of Debian GNU/Linux use the same boot seqence, same filesystem hierarchy, same configuration, same package management, same set of userland tools which can be chosen by sys-admins,... There is no diversity to deal with!
Well... there is version differences, but there is version differences even in BSD, so it shouldn't count.;p
How about the size of user base? Well... Just Debian GNU/Linux beats all BSD summed together hands-down! So don't worry about specializing just in Debian, it is still better than BSD!
If you don't think that "Linux" suffices to pinpoint your knowledge about your OS, just be more specific.
On the other hand, if I understand things correctly, the name sid is special: it is always with unstable. That is, unstable will always be called sid, and new names are given to testing rather than unstable. Which is because sid is the character in toy story which keeps breaking toys, which matches the characteristics of unstable.
You are talking about backporting, etc, using _source_ packages aren't you?
I'm not the poster, so I don't know what he is talking about. But that is possible, and not that difficult. Debian apt uses a file/etc/apt/sources.list to get the sources of the distributions used, and/etc/apt/preferences to store which distribution is preferred to which for each package. So you can add lines to/etc/apt/sources.list to get the unstable or testing list, and then modify/etc/apt/preferences to tell that the stable branch is preferred. See apt_preferences(5) for information.
It's actually funnier than that: apt prefer upgrading than downgrading, so you can configure your system to track stable for most packages, unless you specifically install the unstable version, and for these packages it tracks unstable. Which is exactly what I do with one of my servers: apt and mozilla tracks unstable, everything else tracks stable.
Some has suggested using ext2resize to shrink your partition to half the size, create a reiserfs partition, remove the old one and use reiserfs-resize to enlarge it back to original size. Didn't try that, though.
Safety comes from two things when programming: the language, and the programmer. Even when confronted with stack-based C-style strign, you can easily make things safe if enough care is taken.
The difference is that C and C++ let you choose when you need safety from the language and when you need speed and thus safety from the programmer. Java dictates that you want safety from language and never speed.
> What we need is for dpkg/apt to track the
> stable/frozen/testing branches all at the same
> time and apply the minimum required versions of
> any prerequisite package unless a higher version
> is explicitly requested.
How about "man apt_preferences"? It is in both woody and sid.
in their own meaning of cancer. Except that then their own software is a much bigger cancer. They say that it would be horrible to incorporate GPL software into a proprietary software. But what if you incorporate the proprietary software of another company, say the one of MS (MacroShit)?! For GPL'ed software, at least you can do it. For proprietary software, you'll have a hard time even trying: the source is unavailable.
Slashdot Mirror
Considering that the rest of the whole world had successfully made the change (to some extent), this cannot be a real problem.
> I like how it "started out anonymous"
> (read: Hong Kong was part of the UK) and
> now is becoming "personally trackable"
> (Hong Kong is now part of the PRC).
I tried not to get into the discussion, but this one gets me. No, privacy invasion is not the motivation of personalizing the card. It can be an issue, though. (Yes, I'm a HK citizen.)
Octopus plans to have personalization feature from day 1 of its operation (yes, when UK is in charge of HK), and it is implemented within months after its launch. The motivation for people to personalize their cards is basically for added convenience. E.g., if you lose your card, you can tell the card center about it and make the card a useless plastic. You can link your card with a bank so that recharges are automatically made once the balance gets down to 0. And so on. It is completely optional: if you don't like that, you can instead buy an anonymous card, which is done by pretty much all the people. I doubt any merchant has the restriction that only personalized cards can use their system.
And... Octopus is *not* run by the government here! It is handled by a company, which branch out from MTR (the subway here), to deal with the ever increasing need to shorten time to buy tickets and process the cards before letting passengers to enter their railway systems. There is no linkage between the Octopus companny and the government. If there is any privacy concern, it is due to Octopus, not the government.
Absolute crap. When the software of the whole world is dominated by MS products (at least, many categories of software), and when Taiwan is a major computer equipment user (and also producer), will you remotely think that it has been using only minor $$s in pleasing MS?
And... the "switch" only make sense if you have to pay a lot of money to MS.
What you are talking about is extern, not export. Extern is fully supported by any C++ compiler. Export is not supported by any so far---before this report. Look into The C++ Programming Language (Bjarne Stroustrup) to see what it is. In short, it should allow an uninstantiated template definition to appear in an object file, allowing other object files to somehow use an instantiated version of it without seeing the code again during compilation. Pretty tough requirement indeed.
> I have to find it first, make sure
> I actually found an element, and THEN I can
> erase it. This means I either keep a local
> copy of the iterator, or I call the
> find_first algorithm twice.
I still don't know what you are blowing...
#include
#include
int main() {
vector test;
for (int i = 0; i 10; ++i)
test.push_back(i);
test.erase(test.begin()+5);
test.erase(find(test.begin(), test.end(), 7));
for (int i = 0; i test.size(); ++i)
cout << test[i] << endl;
}
> Just wanted to point out that the STL is
> not, in any way, an OO system. It uses
> classes, but that's as close as it gets.
> There's no use of inheritance, polymorphism
> (in the usual OO sense) or any other
> "typical OO" features.
Everything in STL is compile-time determined. Then there is no need for using OO concepts like polymorphism. I'd rather count generic programming as a concrete part of OO programming, and say that STL used the alternative way to do OO.
> do they really think that mozilla-m18, which is in debian/stable, is more stable or usable than 1.0?! Or that gnome 1.0.55 is better than gnome-1.4.0.
You get the wrong idea about what stable really is. A stable release is something which is more or less *fixed*, i.e., won't change much. When potato is released, there is no gnome-1.4.0, and gnome-1.2 is just out which is not in production quality yet. So not including them is the right choice.
We will see similar choice made in this stable release. Apache 2 is not stable enough, so 1.3 is shipped. xFree 4.2 is not there long enough yet, so we have XFree 4.1. This is no fault of Debian at all. Stable is for thnigs that "have to be up with no incompatible changes to programs" for a long time. If you don't need such a beast, you can always ignore stable (as I do) and always use testing or unstable.
If you don't use Debian, try to use it. If you are still thinking that everybody should use stable, try the freezed and testing release, or even the unstable release.
Debian might not be the norm in calling something stable only when it mostly won't change. But if you think that you should be using software that are more recent than a year old, then you have no business using Debian stable. It is simply not for that purpose. Use testing or unstable instead.
Is it just me who experience this? On my notebook LCD, I see exactly the same problem described here: at lower resolution, the display is ugly. But on my LCD panel I used on my desktop, it is much less a problem. It seems that the panel does anti-aliasing, but I'm not sure. Of course, not as good as a CRT. But still, much better, to the point that I don't really care.
Just hope that Transmeta will still be living well by the time when OLED display is out there. :)
> Secondary it shows that he isn't interrested in any workable bussinessmodel for free software since he hails the good in people swapping software without paying the creator.
Think again. Remember that for many years RMS pay himself by writing free software. His opposition to business models just go to the extend that "if the business model ruin our freedom, than replace it as soon as we can".
So he call for trashing the current business model. But he doesn't say that all business models are bad.
With the constraint that the contest should be timed, should be about coding, should be language neutral (to different languages), should be reasonable to be completed within one day, and should be judged objectively rather than subjectively, I see no alternative. Will you care to suggest some?
More importantly, whatever your contest rules, there are some skills needed in programming that is not tested at all. Will you say that it "is not a programming contest" because they don't test the skills in those areas?
Come on. A contest is there to give motivations to work hard, and to gather people so that different bright people can meet each other and share ideas. There is no ``ultimate game rule'' that tests everything. And that is not the objective of the contest.
Solution: of course, work it out yourselves. :)
Penalty: each correctly answer carries a penalty, which is the amount of time elapsed from the beginning of the contest until the correct answer is received, plus 20 minutes times the number of time the same question is submitted with an incorrect answer. So the penalty listed is simply the total penalty (less is better).
Most people who has such a feeling of the contest should go back and check the question again, and think about what will happen if the input is really as large as suggested by the question. You can solve it the Prolog way and use backtracking and so on to find the solution. But the judge will simply answer "time limit exceeded" to your submission of answers, nothing else.
Right. Remember that the problem set is done by 3 contestents, not 1; and all of them share one copy of the questions. It is very frustrating having to read half of the question and wait for your peer to finish reading his part and give you the other half.
> Just as C++ will never replace C (in the
> forseeable future), because C++ has its own levels
> of indirection and safety which slow it down
> (RTTI, virtual tables, etc)
I can't agree with this. RTTI, virtual tables, etc., never shows up until you actually use the feature. If you decide it that way, you can always use the C library subpart of the language and forget about dynamic binding. RTTI is actually rarely used even in the standard library of C++.
On the other hand, if you need the capability, you can't do it better in C. If you write a C language object system, chances are that you will get worse performance, rather than better performance, than a corresponding C++ system, because the developers of the C++ compilers have really put a lot of efforts into supporting the OO concepts in language level.
But your main point is still correct: C++ contains some features that will make it slower and more bulky. Basically, exceptions. If you go into the compiled assembly generated by any recent C++ compiler, you'll find that 20% of the code is generated only to get exception handling correct. While C++ designers will tell you that it won't cause performance problems, in reality it clutter the cache, create a few more branches to fool branch prediction hardware, etc.
But most C++ programs are slow for a completely different reason: the people are more concerned about other things like adding features and maintaining software.
That said, the most serious problem of C++ against C is not performance issue. It is the learning curve issue: C is a small language. You can learn it in a couple of days, and start your venture into each of the library. C++ is a huge language: the language specification goes over 1000 pages, and you'd like to understand the whole standard library before actually working on any real project. That makes it very unattractive for people who don't have their background in C.
There can be only one way for human evolution to really be "over": the natural selection won't work anymore. In other words, everybody survive and can have equal chance to give offsprings, regardness of the shortcoming of the individual. The gene pool will no longer be able to benefit from natural selection.
Unluckily, the current medical advances are really leading us to this sequel.
> Compare with the Linux world, where you have a
... There is no diversity to deal with!
;p
> bunch of different distros that many people
> pretend are all the same OS (in spite of the fact
> that file systems are arranged differently, boot
> sequences are different, configuration is
> different, package management is different,
> userland tools are often different, etc.) because
> they happen to use the same kernel.
That makes it extremely easy... just use <add-your-dist-name-here>Linux! I use Debian GNU/Linux, and every installation of Debian GNU/Linux use the same boot seqence, same filesystem hierarchy, same configuration, same package management, same set of userland tools which can be chosen by sys-admins,
Well... there is version differences, but there is version differences even in BSD, so it shouldn't count.
How about the size of user base? Well... Just Debian GNU/Linux beats all BSD summed together hands-down! So don't worry about specializing just in Debian, it is still better than BSD!
If you don't think that "Linux" suffices to pinpoint your knowledge about your OS, just be more specific.
On the other hand, if I understand things correctly, the name sid is special: it is always with unstable. That is, unstable will always be called sid, and new names are given to testing rather than unstable. Which is because sid is the character in toy story which keeps breaking toys, which matches the characteristics of unstable.
I'm not the poster, so I don't know what he is talking about. But that is possible, and not that difficult. Debian apt uses a file /etc/apt/sources.list to get the sources of the distributions used, and /etc/apt/preferences to store which distribution is preferred to which for each package. So you can add lines to /etc/apt/sources.list to get the unstable or testing list, and then modify /etc/apt/preferences to tell that the stable branch is preferred. See apt_preferences(5) for information.
It's actually funnier than that: apt prefer upgrading than downgrading, so you can configure your system to track stable for most packages, unless you specifically install the unstable version, and for these packages it tracks unstable. Which is exactly what I do with one of my servers: apt and mozilla tracks unstable, everything else tracks stable.
Some has suggested using ext2resize to shrink your partition to half the size, create a reiserfs partition, remove the old one and use reiserfs-resize to enlarge it back to original size. Didn't try that, though.
Safety comes from two things when programming: the language, and the programmer. Even when confronted with stack-based C-style strign, you can easily make things safe if enough care is taken.
The difference is that C and C++ let you choose when you need safety from the language and when you need speed and thus safety from the programmer. Java dictates that you want safety from language and never speed.
> What we need is for dpkg/apt to track the
> stable/frozen/testing branches all at the same
> time and apply the minimum required versions of
> any prerequisite package unless a higher version
> is explicitly requested.
How about "man apt_preferences"? It is in both woody and sid.
in their own meaning of cancer. Except that then their own software is a much bigger cancer. They say that it would be horrible to incorporate GPL software into a proprietary software. But what if you incorporate the proprietary software of another company, say the one of MS (MacroShit)?! For GPL'ed software, at least you can do it. For proprietary software, you'll have a hard time even trying: the source is unavailable.