Granted, the *nix way of doing things isn't perfect, but at least it had homedirectories(!) with.apprc settings ages ago.
Yeah, like we really needed thousands of.whateverc files in our home directory. Not to mention countless environment variables, sometimes unpleasantly clashing with each other.
I like to keep things *clean*. I don't want endless weird-named files in my ~/, not even if ls conveniently hides them. Why isn't this in ~/etc/? I know - hysterical raisins, of course, but surely it isn't too late to change is it?
And the same thing goes for environment variables. Especially those with short, common names - if your application needs configuration, for heavens sake use a configuration file. And put it in ~/etc/ while you are at it...
Clearly, you have a point. It's obvious that Gates couild be lying about this. But there's no question that there _could_ be Microsoft IP in Open Source, simply because there is a lot of Microsoft IP out there and a lot of source in which it could be embedded.
It is true there is no legal recourse *yet*, but we now know beyond doubt that a highly placed russian government official is aware of the problem. This raises the hope that a law against spam could be in the works too.
Of course, being the russian government, they do have other options, like sending in the special forces for example. It wouldn't have to cost them anything - spammers are not likely to fight back, and I'm sure people would pay to see footage of a swarm of Hinds obliterating a spammers hideout;-)
The major difference, I think, is that in the past a programmer would write an engine *first* and then start thinking about how to write a game around it. This way we ended up with some pretty unique gaming concepts.
These days games are almost exclusively written the other way around: people think of the "skin" first (i.e. what it looks like, what it plays like) and then re-use an existing (possibly commercial) engine.
The first approach allows for innovation and inspiration; the second approach leads to an endless string of me-too titles. Before I'm accused of having rose-tinted glasses, I also believe the amount of trashy games is constant throughout the space-time continuum...
As a form of proof I offer this link; these games were written around an engine rather than the other way around, and while they are by no means finished products, you can see inspiration shine through.
System Shock 3 is unfortunately rather unlikely, since apparently the rights to the game are split over so many different parties. If that's true, it is a major shame because SS2 is crying for a sequel with modern graphics. Hell, I'd buy a remake (using the Doom 3 engine, perhaps) in a second.
SS2 is easily the scariest game I've ever played. One of the elements that helps produce that fear is the sense of being alone, the last human survivor on that once-proud outpost of humanity, surrounded by the twisted shapes of your former fellow crew members. So, the log system works well - you are not supposed to make contact with other humans.
That is not to say there aren't any - there are at least two other survivors (of which you catch a brief glimpse when they escape the ship using an escape pod) and there's of course dr. Janice Polito who is guiding you (by radio) through the first four levels.
But still - the logs tell the sad stories of the crew going about their daily business, unaware that they only have hours to live; the stupid decisions by the various commanding officers, leading to the disaster to follow; the desperate attempts to regain control of the ship, and later to simply survive. It's gripping stuff.
For some reason my copy of SS2 no longer works - the graphics completely fall apart when enemies fire at me, and in certain rooms. This didn't happen the first time I played it, but I've installed a new DirectX and a new Detonator driver since then. Has anybody else experienced this? Is there a solution for it? I'd love to play this game again, especially using those new models from SS2: Rebirth.
As for the other games you mention, I agree about Thief (it seems just too tiresome to play), Deus Ex is absolutely fantastic, and the head crabs from HL seem rather inspired by the movie "Alien";-)
Not only was ARexx great for scripting any ARexx-aware application, but more importantly, it could be used to connect any number of such applications together (and not just with a simple pipe of data from one app to another either - a script could be performing some complex operation by tying together two, three, or more apps as a single unified whole).
I always get the feeling that people who have never used an Amiga do not fully appreciate the kind of power that ARexx offers, so let me try to provide a slightly longer description.
To make an application ARexx-aware, it must open a named message port. That is Amiga-speak for providing an interface to which the ARexx scripting engine can connect.
To connect to an application, the script uses the "command " function. The script engine searches the system for the named port, and if found will connect to it. When the script contains a function call the script engine does not recognize (i.e. one that is not a built in) it will send a message to the port, asking the application to process the function instead. Thus, an ARexx script can interact with an application while it is running, which makes it fundamentally different from, say, Perl, which can start external applications but will not typically interact with them while they are active (I'm not saying "cannot", just that it doesn't happen very often).
ARexx scripts can interact with multiple application by switching message port as required, simply by executing the "command " again and again.
Over the years ARexx has played an enabling role for the Amiga, since it made possible all those applications that no software author ever considered. There were scripts for special text effects in a word processor; for automatically playing chess over email (by tying together a newsreader and a chess program); for classifying incoming email; for automatically cropping and printing scanned images (by tying together a scanner program and an image manipulation tool); and a thousand other things. For me, the killer apps were the ARexx-aware text editor Turbo Text (which I used to perform a couple of minor miracles) and the ARexx-aware file utility (Norton-like) Disk Master.
You could build applications that do all these things, but the point is that ARexx (as a design philosophy) is a very cheap way to design for those future applications that you as a software author are not yet aware of. It would make an excellent fit to the UNIX philosophy of having many small components working together.
Our company template uses numbered tables, but OO displays numbers that are different from MSO. For example, if there is a table "1" in section "5.1" than in MSO the number is "5.1.1" but in OO it becomes "5.1-5.1.1" which is rather ugly.
In the table of contents, there is a bit of space between the section number and title in MSO, but OO concatenates the number and title, which also looks ugly.
We also spotted an empty chapter 1 before the actual text started, which was not present in the MSO interpretation of the document. This also means all the chapter numbers changed, which we really don't like.
Finally, in the header the document title suddenly popped up twice whereas MSO just displays it once. There could be a hidden field there that gets displayed anyway by OO, I haven't checked yet.
Before you ask, I haven't reported any of these problems yet. Don't shoot, they were only discovered yesterday...
And don't take all this as whining: I am very pleased by the way OO is progressing, and I want to help making it a little better by reporting these errors.
Or any other OS? Any webbrowser? Any database product?
I have to admit I'm unsure how to read that "enable 3rd parties" bit. Does that mean that it must enable someone else to upload to my machine? Or does it mean that it must enable me to download stuff from elsewhere?
No. That's because "++" can only operate on an l-value. An l-value is something that can appear on the left-hand side of an assignment (in other words, it must be something that can be assigned to). "--input" is not an l-value, and thus cannot be used with "++".
This is also true for other combinations of "++" and "--" operating on a single variable (like "(a++)++") - the language simply does not allow them.
There are a couple of further complications too. For example, anything that takes the form of "a = a++" or "b = a++ + a++" will produce unpredictable results (that means that some compilers might do what you expect, but others could do something else). If you want to know more about this, try searching for "sequence point".
One good place to look is here, especially in chapter 3 which deals with these issues.
Sorry, but that's not the same since it also increases "input" by 1, which is not what the original statement did.
As for the other poster, "++variable--" is illegal in C++. And let's all be grateful for that - I'd hate having to kill those colleagues that used such "gems" in their code...
Actually "result += input" should be read as "result = result + input", rather than "result = input + 1". There is no shorthand form for that last statement.
As for readability, I find that the actual symbol used to describe a notation doesn't change complexity much. The real complexity is always in the algorithm (in other words, in what the program is trying to accomplish). How you write that down may take a few more or less characters depending on language but it does not fundamentally change the difficulty of the problem.
That's not to say that I do not appreciate some useful tools like the C++ (I do hardcore C++ for my work so forgive me for being C++-centric) map, or string class, or even the concept of OO, and for sure, some languages tend to allow more obfuscation than others, but none of this alters the fundamental difficulty of programming (which is why you occasionally see the term "syntactic sugar").
Of course that derogatory term is most often used by misguided academics who fail to realize that those useful tools help a great deal in scaling from a one-page example to a million line working system...
Harry Potter is the star of an utterly overhyped fantasy series aimed at children. He is an orphan magician who goes to a magical school somewhere in the UK, and gets into all sorts of magical adventures with a couple of friends. So far, five volumes have been released, each covering one year of Harry's life. The books are not extremely bad or something, but certainly not worth the incredible hype surrounding them.
Congrats on missing them, you'd better keep your location secret so it cannot be traced by the Potter-mafia... Posting as AC makes for a good start;-)
If you are simply looking for good fantasy, I'd recommend "The View from the Mirror" by Ian Irvine, or the "Chronicles of Thomas Covenant" by Stephen Donaldson, or "A Song of Ice and Fire" by George Martin, or "The Initiate Brother" by Sean Russell, or "The Sarantine Mosaic" by Guy Gavriel Kay. I found myself immensely enjoying all of these, as well as other books by those same authors. Having said that, these are all aimed at a more mature audience.
Organize my MP3 collection with FS metadata and lose it all when I try to move to another FS? What is he thinking? Is he thinking at all?
AmigaOS had a single meta-data field in its various filesystems, called the "file comment" or "file note". While it was clearly far less powerful than what is proposed here, it was incredibly useful for lots of purposes.
The problem you describe, losing the meta-data when moving to another filesystem, was non-existant on the Amiga for two reasons: all filesystems implemented filenotes, and LhA (the Amiga-equivalent of tar/gzip) saved and restored filenotes. A similar solution can be used for Linux (tar and gzip are getting very long in the tooth anyway).
I would love to see proper meta-data fields implemented throughout the system, especially if it is pervasive throughout the OS (so I can search for specific values from the desktop, for example).
Finally, I believe BeOS had something similar, and users of that filesystem also tend to be enthousiastic about it. All this doesn't mean the current solution is necessarily the best one, of course - just that the concept is pretty damn useful.
Re:Text import in 1.0.x was slow
on
Gnumeric Turns 5
·
· Score: 1
1) That's the definition of "change" I was looking for;-)
2) And again, yes that's the one I want. Thanks for the answers. The company I work for has become really open-minded towards Open Source lately (Open Office on every in-house desktop, Linux projects at customer sites, etc.) and a good spreadsheet (with these specific capabilities) is just the thing we need to solve a few problems.
Re:Text import in 1.0.x was slow
on
Gnumeric Turns 5
·
· Score: 1
Our goal is to produce the best damn spreadsheet around. This is still version 1.1, 2.0 (extend) and 3.0 (extinguish) aren't due for a while yet.
I commend you on your sense of ambition.
While I'm posting anyway, is there a method in Gnumeric to change a spreadsheet under control of another program in real time? I have a large body of data that is being gathered continuously and that my users like to look at in a spreadsheet while their process is running - does Gnumeric support such a thing? (I looked, but couldn't find it on the "features" page)
Unrelated to the previous question, does Gnumeric support timestamps with a resolution of less than a second?
Re:What happened to Robert Jordan?
on
A Game of Thrones
·
· Score: 1
I liked the first two or three books from The Wheel of Time, and hated the rest for just rambling on and on and on and on... By comparison, although A Game of Thrones is similarly large in scope, it moves at a much faster pace. So far very little rambling has been detected, and since the author appears to actually *plan* his work, there is some ground for optimism for the future titles. Go on, try it.
For years people have been saying there is no money to be made in open source. Why not build a company based on the idea of selling support to companies?
There are plenty good people out there right now who don't have a job. Set up a company with a couple of other knowledgeable people (you don't want to man the 24x7 phone on your own) and start selling Linux support. It will solve your own problem and the Linux acceptance problem.
In good slashdot style:
1. Start a company selling Linux support.
2. ***ADVERTIZE, DAMNIT!***
3. Profit!
(Sorry for violating slashdot etiquette by filling in that second, elusive step but this is a serious post...)
I agree with that, but if the manned space program was stopped "because it is not interesting anymore" it may never come back. The funds will be taken away from NASA and used for something else, and pretty soon the knowledge will die out as old engineers leave and new ones are not trained or hired.
You may not think of this as a disaster, but if we close the window we may never have another opportunity to explore space again. There will always be a slightly more pressing concern, some war to be fought or some local problem to be solved, after which we may start thinking about space again. Only that moment will never come, and reassembling the knowledge will become harder and harder.
Meanwhile, the upstart super powers will have no qualms about putting men in space (even if it is only to give you the finger while they pass overhead). They have something to prove, and rest assured they will do that. And I'm sure they will gracefully allow american astronauts to fly on their rockets, to give you a little toe in their space program (assuming that astronaut learns chinese or french or whatever the appropriate language will be). At that point the USA may still be a superpower, but like the ex-USSR today it will be considered a second-rate superpower.
Do you really want your best engineers to migrate to China and India and Europe because that's where their talents will be appreciated and used? Are you willing to give up on the notion of the USA being a superpower, and letting other countries take over?
American enterprise can already go where it will in space. What you are actually asking for is for these 'private' enterprises to be sponsored by the public, isn't it?
These continued calls to abandon NASA and "let private enterprises take over" sicken me. If private enterprises want to take over they can do so, without the need to abandon NASA first. The simple fact is that private enterprises have done exactly *nothing* that could be called "space exploration" so far. There are no private space shuttles, nor any private expendable launchers that can take any meaningful payload to a meaningful orbit.
As soon as private enterprises start offering the same sort of services as NASA does I agree it may be time to shut it down. Until then there is simply no alternative. Shutting it down prematurely will simply mean that the USA loses access to space. If you believe that's acceptable, hey, fine with me. The Russians, Europeans, Chinese, and Indians will love you for it.
Assuming SCO wins and IBM is ordered to remove the offending code, what would happen if Linus said "no" to IBM's new kernel patches?
IBM does not carry responsibility for Linux, and has no direct ability to change it. If they are forced to change it but cannot, how will this case proceed?
I didn't even realize there was a rating "5, troll"... Anyway, how precisely is this a troll? The guy is absolutely right. The first n-1 pages of the article describe the evolution and functioning of the various Windows filesystems (and badly, at that), and the last page just jumps to the conclusion that this new FS is going to be much better than anything that came before it. There is no description of features or functionality to accompany this claim; in fact the only shred of evidence that is presented is the size of the module that implements the FS: 20MB. Which is ridiculously large for a FS, but that's a topic for another story...
Slashdot Mirror
Yeah, like we really needed thousands of .whateverc files in our home directory. Not to mention countless environment variables, sometimes unpleasantly clashing with each other.
I like to keep things *clean*. I don't want endless weird-named files in my ~/, not even if ls conveniently hides them. Why isn't this in ~/etc/? I know - hysterical raisins, of course, but surely it isn't too late to change is it?
And the same thing goes for environment variables. Especially those with short, common names - if your application needs configuration, for heavens sake use a configuration file. And put it in ~/etc/ while you are at it...
Clearly, you have a point. It's obvious that Gates couild be lying about this. But there's no question that there _could_ be Microsoft IP in Open Source, simply because there is a lot of Microsoft IP out there and a lot of source in which it could be embedded.
Of course, being the russian government, they do have other options, like sending in the special forces for example. It wouldn't have to cost them anything - spammers are not likely to fight back, and I'm sure people would pay to see footage of a swarm of Hinds obliterating a spammers hideout ;-)
Like the author of "Tanja Grotter" was sued by the author of "Harry Potter"? And lost because the plotlines were too similar? See here for example...
Actually I'm more inclined that this was a case of seriously fucked up justice, but it goes to show that it _does_ happen.
They also support United Linux. Look in the Oracle Linux FAQ, here for example.
These days games are almost exclusively written the other way around: people think of the "skin" first (i.e. what it looks like, what it plays like) and then re-use an existing (possibly commercial) engine.
The first approach allows for innovation and inspiration; the second approach leads to an endless string of me-too titles. Before I'm accused of having rose-tinted glasses, I also believe the amount of trashy games is constant throughout the space-time continuum...
As a form of proof I offer this link; these games were written around an engine rather than the other way around, and while they are by no means finished products, you can see inspiration shine through.
That is not to say there aren't any - there are at least two other survivors (of which you catch a brief glimpse when they escape the ship using an escape pod) and there's of course dr. Janice Polito who is guiding you (by radio) through the first four levels.
But still - the logs tell the sad stories of the crew going about their daily business, unaware that they only have hours to live; the stupid decisions by the various commanding officers, leading to the disaster to follow; the desperate attempts to regain control of the ship, and later to simply survive. It's gripping stuff.
For some reason my copy of SS2 no longer works - the graphics completely fall apart when enemies fire at me, and in certain rooms. This didn't happen the first time I played it, but I've installed a new DirectX and a new Detonator driver since then. Has anybody else experienced this? Is there a solution for it? I'd love to play this game again, especially using those new models from SS2: Rebirth.
As for the other games you mention, I agree about Thief (it seems just too tiresome to play), Deus Ex is absolutely fantastic, and the head crabs from HL seem rather inspired by the movie "Alien" ;-)
I always get the feeling that people who have never used an Amiga do not fully appreciate the kind of power that ARexx offers, so let me try to provide a slightly longer description.
To make an application ARexx-aware, it must open a named message port. That is Amiga-speak for providing an interface to which the ARexx scripting engine can connect.
To connect to an application, the script uses the "command " function. The script engine searches the system for the named port, and if found will connect to it. When the script contains a function call the script engine does not recognize (i.e. one that is not a built in) it will send a message to the port, asking the application to process the function instead. Thus, an ARexx script can interact with an application while it is running, which makes it fundamentally different from, say, Perl, which can start external applications but will not typically interact with them while they are active (I'm not saying "cannot", just that it doesn't happen very often).
ARexx scripts can interact with multiple application by switching message port as required, simply by executing the "command " again and again.
Over the years ARexx has played an enabling role for the Amiga, since it made possible all those applications that no software author ever considered. There were scripts for special text effects in a word processor; for automatically playing chess over email (by tying together a newsreader and a chess program); for classifying incoming email; for automatically cropping and printing scanned images (by tying together a scanner program and an image manipulation tool); and a thousand other things. For me, the killer apps were the ARexx-aware text editor Turbo Text (which I used to perform a couple of minor miracles) and the ARexx-aware file utility (Norton-like) Disk Master.
You could build applications that do all these things, but the point is that ARexx (as a design philosophy) is a very cheap way to design for those future applications that you as a software author are not yet aware of. It would make an excellent fit to the UNIX philosophy of having many small components working together.
Is it time for LRexx?
Let me rephrase that: it is time for LRexx.
In the table of contents, there is a bit of space between the section number and title in MSO, but OO concatenates the number and title, which also looks ugly.
We also spotted an empty chapter 1 before the actual text started, which was not present in the MSO interpretation of the document. This also means all the chapter numbers changed, which we really don't like.
Finally, in the header the document title suddenly popped up twice whereas MSO just displays it once. There could be a hidden field there that gets displayed anyway by OO, I haven't checked yet.
Before you ask, I haven't reported any of these problems yet. Don't shoot, they were only discovered yesterday...
And don't take all this as whining: I am very pleased by the way OO is progressing, and I want to help making it a little better by reporting these errors.
I have to admit I'm unsure how to read that "enable 3rd parties" bit. Does that mean that it must enable someone else to upload to my machine? Or does it mean that it must enable me to download stuff from elsewhere?
This is also true for other combinations of "++" and "--" operating on a single variable (like "(a++)++") - the language simply does not allow them.
There are a couple of further complications too. For example, anything that takes the form of "a = a++" or "b = a++ + a++" will produce unpredictable results (that means that some compilers might do what you expect, but others could do something else). If you want to know more about this, try searching for "sequence point".
One good place to look is here, especially in chapter 3 which deals with these issues.
As for the other poster, "++variable--" is illegal in C++. And let's all be grateful for that - I'd hate having to kill those colleagues that used such "gems" in their code...
As for readability, I find that the actual symbol used to describe a notation doesn't change complexity much. The real complexity is always in the algorithm (in other words, in what the program is trying to accomplish). How you write that down may take a few more or less characters depending on language but it does not fundamentally change the difficulty of the problem.
That's not to say that I do not appreciate some useful tools like the C++ (I do hardcore C++ for my work so forgive me for being C++-centric) map, or string class, or even the concept of OO, and for sure, some languages tend to allow more obfuscation than others, but none of this alters the fundamental difficulty of programming (which is why you occasionally see the term "syntactic sugar").
Of course that derogatory term is most often used by misguided academics who fail to realize that those useful tools help a great deal in scaling from a one-page example to a million line working system...
Congrats on missing them, you'd better keep your location secret so it cannot be traced by the Potter-mafia... Posting as AC makes for a good start ;-)
If you are simply looking for good fantasy, I'd recommend "The View from the Mirror" by Ian Irvine, or the "Chronicles of Thomas Covenant" by Stephen Donaldson, or "A Song of Ice and Fire" by George Martin, or "The Initiate Brother" by Sean Russell, or "The Sarantine Mosaic" by Guy Gavriel Kay. I found myself immensely enjoying all of these, as well as other books by those same authors. Having said that, these are all aimed at a more mature audience.
AmigaOS had a single meta-data field in its various filesystems, called the "file comment" or "file note". While it was clearly far less powerful than what is proposed here, it was incredibly useful for lots of purposes.
The problem you describe, losing the meta-data when moving to another filesystem, was non-existant on the Amiga for two reasons: all filesystems implemented filenotes, and LhA (the Amiga-equivalent of tar/gzip) saved and restored filenotes. A similar solution can be used for Linux (tar and gzip are getting very long in the tooth anyway).
I would love to see proper meta-data fields implemented throughout the system, especially if it is pervasive throughout the OS (so I can search for specific values from the desktop, for example).
Finally, I believe BeOS had something similar, and users of that filesystem also tend to be enthousiastic about it. All this doesn't mean the current solution is necessarily the best one, of course - just that the concept is pretty damn useful.
2) And again, yes that's the one I want. Thanks for the answers. The company I work for has become really open-minded towards Open Source lately (Open Office on every in-house desktop, Linux projects at customer sites, etc.) and a good spreadsheet (with these specific capabilities) is just the thing we need to solve a few problems.
I commend you on your sense of ambition.
While I'm posting anyway, is there a method in Gnumeric to change a spreadsheet under control of another program in real time? I have a large body of data that is being gathered continuously and that my users like to look at in a spreadsheet while their process is running - does Gnumeric support such a thing? (I looked, but couldn't find it on the "features" page)
Unrelated to the previous question, does Gnumeric support timestamps with a resolution of less than a second?
I liked the first two or three books from The Wheel of Time, and hated the rest for just rambling on and on and on and on... By comparison, although A Game of Thrones is similarly large in scope, it moves at a much faster pace. So far very little rambling has been detected, and since the author appears to actually *plan* his work, there is some ground for optimism for the future titles. Go on, try it.
There are plenty good people out there right now who don't have a job. Set up a company with a couple of other knowledgeable people (you don't want to man the 24x7 phone on your own) and start selling Linux support. It will solve your own problem and the Linux acceptance problem.
In good slashdot style:
1. Start a company selling Linux support.
2. ***ADVERTIZE, DAMNIT!***
3. Profit!
(Sorry for violating slashdot etiquette by filling in that second, elusive step but this is a serious post...)
You may not think of this as a disaster, but if we close the window we may never have another opportunity to explore space again. There will always be a slightly more pressing concern, some war to be fought or some local problem to be solved, after which we may start thinking about space again. Only that moment will never come, and reassembling the knowledge will become harder and harder.
Meanwhile, the upstart super powers will have no qualms about putting men in space (even if it is only to give you the finger while they pass overhead). They have something to prove, and rest assured they will do that. And I'm sure they will gracefully allow american astronauts to fly on their rockets, to give you a little toe in their space program (assuming that astronaut learns chinese or french or whatever the appropriate language will be). At that point the USA may still be a superpower, but like the ex-USSR today it will be considered a second-rate superpower.
Do you really want your best engineers to migrate to China and India and Europe because that's where their talents will be appreciated and used? Are you willing to give up on the notion of the USA being a superpower, and letting other countries take over?
These continued calls to abandon NASA and "let private enterprises take over" sicken me. If private enterprises want to take over they can do so, without the need to abandon NASA first. The simple fact is that private enterprises have done exactly *nothing* that could be called "space exploration" so far. There are no private space shuttles, nor any private expendable launchers that can take any meaningful payload to a meaningful orbit.
As soon as private enterprises start offering the same sort of services as NASA does I agree it may be time to shut it down. Until then there is simply no alternative. Shutting it down prematurely will simply mean that the USA loses access to space. If you believe that's acceptable, hey, fine with me. The Russians, Europeans, Chinese, and Indians will love you for it.
IBM does not carry responsibility for Linux, and has no direct ability to change it. If they are forced to change it but cannot, how will this case proceed?
- right
that you have to own the ROMs before you are allowed to play them in this manner, the sad fact is that most people just don't- care
about this issue and just play the ROMs. Just goes to show that there is a difference between law and morality...I didn't even realize there was a rating "5, troll"... Anyway, how precisely is this a troll? The guy is absolutely right. The first n-1 pages of the article describe the evolution and functioning of the various Windows filesystems (and badly, at that), and the last page just jumps to the conclusion that this new FS is going to be much better than anything that came before it. There is no description of features or functionality to accompany this claim; in fact the only shred of evidence that is presented is the size of the module that implements the FS: 20MB. Which is ridiculously large for a FS, but that's a topic for another story...