I'll second al broccoli's response - the parent post is talking absolute crap.
I, we - the team I work in, have recently finished a migration of SAP4.6 on windows to SAP4.7 on solaris. The windows installation was formerly the largest single installation of SAP on windows in western europe, and it had severe scaling and support problems.
I also know from some years back, this having been discussed in an interview I attended, that Unilever, who before they switched held the crown for the largest installation of SAP on windows, switched to Digital UNIX (before it was Tru64) for exactly the same reason.
SAP on any OS has a hefty hardware requirement list. In addition, to my mind, they make some stupid recommendations about memory. Viz - for our setup we have multiple V440 and V490 suns with 16GB memory, and SAP want (and, on some servers, having hit this problem, *need*) THREE or FOUR TIMES that in swap. I would have suggested either less databases on each machine, or more memory (not sure if 16GB is the physical limit for V440/490) or just bigger machines, but then it's not my job to spec these things.
Anyway, having vast quantities of swap actively used as working memory may have contributed to the instability of a SAP system on linux, if as someone suggested that VM on linux is currently not as good as it could be.
Firstly, thank you for a much more informative reply than your initial complaint.
I still have the impression that you dont have a lot of patience, and that you're so comfortable with BT that anything which doesnt do things in exactly the same way is not good enough. Please remember that neither is good reason to criticise ed2k/emule. Clearly, BT suits your needs much better than ed2k, but that doesnt make BT "right" and ed2k "wrong".
Incidentally, please excuse my aggressive tone where I refute some of your points - it is not intended to offend. --- I am still not convinced that emule is confusing or difficult to set up. You have not actually identified *what* the problems are. Yes, elitist attitudes are a problem, and yes, it is good that you work to develop autonomic software, but this is just arm-waving in relation to emule, not a proper complaint.
Also, the very popularity of the ed2k network provides empirical evidence that several million people *can* get it working rather well. That BT is *more* simple does not imply emule is hard to set up, only *comparitively* harder.
Servers do come and go on the list. However, as you have not eliminated, nor it seems evaluated the liklihood, that many of those servers are unreliable, point the blame at the emule client is not justified. There are a number of reliable, long-standing servers, and they tend to have large numbers of connected users. A large number of servers appear with just one user, suggesting to me that they are set up "because it's fun". In any case, emule handles variable server presence - you should examine the static server list and automatic reconnection options.
Bootstrapping: emule's kad implementation avoids *all* servers by design; it eliminates the dependence on servers, which is the goal of a fully p2p network.
Azureus' approach may well be more *convenient* to you logging in. "Effective" depends on point of view and stated purpose (such as there is one, of the network).
The one you want is gruk.org, #8 on the list. The #1 entry *does* have some working servers in its list, though the list is old. Many links have server.met files which you'd need to put in the emule config directory - not intuitive, I grant you, though the guide at emule-project.net does describe this, I believe.
With regard to BT's apparent speed over emule, you might want to try the following: set BT to download something large and popular, say, an entire season of Babylon 5, or Galactica. Do the same on ed2k. I grant you, ed2k may take an hour or two to get started, but in my experience it can easily saturate a 2MBit line. I'm sure BT will do the same.
emule is perfectly capable of moving large amounts of data in a short time. However, it may not do it in the order *you* want. If that's not convenient, fine, stick with BT, but *do not* criticise emule for not being the tool it is not designed to be.
(side note: new releases: TV prog shown friday, emule download complete by saturday evening - not as fast as BT, but then I have patience;)
Searching: yes, ed2k's search is, well let's say erratic. Because it *has* a search, the equivalents of piratebay are not required, but an optional extra (so the quality of such sites are likely to be reduced - though realworld and anidb.info are, as far as I can see, excellent). Realworld is under legal attack, just like Loki and Suprnova were - again, you cannot criticise emule for suffering from the same legal problems as BT.
Useless peers: you can't tell whether a client is *refusing* to upload to you unless you can see its client list (which you can't). All you can see is your position on the queue and yes, sometimes you can be #1 in the queue and drop down without getting anything - caused by re-ask
eMule does not pander to the "I want it NOW!" attitude.
I'm not going to answer every point, because there's far too much, and it strikes me that you know very little about emule/ed2k and haven't tried very hard to find out.
1) Setup and use is much too confusing.
Confusing? How? Did you examine the guide at emule-project.net? Or are you just assuming that because BitTorrent (BT) does everything for you emule will do the same?
...eMule tries to manage a list of servers, and doesn't seem to do a very good job of it.
Substantiate and justify this please. (Lists of servers are largely irrelevant. One server with a large number of users is enough. See +++ below.)
2) It doesn't "just work". Getting your client to connect to the kademlia network is a nightmare,...
Kademlia (kad) is fully p2p, no servers. You need previously-known contacts to connect. Every time you download or upload a file part from/to a client which *is* on kad, your client records that person as a contact. When you start off, you have no contacts, you need to be patient, and in any case you dont have to use kad.
For the server method, you only *need* one server to start - your client will learn about other servers +++ from other clients as you exchange file parts. You can *google* for server lists too, if you must. There may even be a server list linked from emule-project.net
3) Downloads are slow.
Downloads are not *instant*. Yes, there is the queuing system. Please explain to me why you should jump ahead of everyone else who is already waiting for a file? The source's bandwidth is a limited resource. emule slices by time, I'm guessing BT slices by speed. Incidentally, if you are downloading a large, popular file, receiving several parts are once, emule can really eat bandwidth. It comes down to the number of people sharing a file, ultimately.
4) It is hard to search.
At least it *has* a built-in search facility. Different search results are caused by the nature of the search mechanism (more so for kad). The search is *not* an index (unlike, say, google). Also, emule tends not to do a full search if it quickly finds >50 unique matches. And the total number of unique matches is limited to 300. With BT, you're searching (fixed) index sites. emule has equivalents, such as the-realworld, osloskop, osiolek, if you care to look.
As a side note, the very presence of a file in an emule-visible place on your computer means it can be found by someone else through a search, by just the filename. By contrast with BT (I believe) if someone doesnt have a.torrent for a file AND publish it via some *non-BT* method, such as a website, you cant ever get at a file being shared. That is, ed2k provides a way to get the filehash (torrent) without *having* the filehash.
5) eMule "swarms" have tons of useless peers....In a BitTorrent swarm, EVERYBODY is uploading
You cant upload before you have parts of the file *to* upload, true of emule and BT. And with emule you *must* share/upload those parts you already have. The emule credit system promotes (through queues) those who upload (RTFM for details). You *could* modify a client to *not* upload, but it would hurt you, because you'll sit in queues for longer. Incidentally, emule doesnt have "swarms" per se. Overnet did.
Until *you* have parts of a file to upload, *you* are a "useless peer". "Useless peers" cease to be useless when they acquire file parts.
...seeding too many files...I will give eMule one thing, it DOES have a lot of rare stuff.
You dont think these two are connected? People *still* sharing things they got >6 months ago while downloading other things? (There is also no "seeding" per se on emule.)
I spent a week downloading a 90MB file,...
(sarcasm) You novice! (/sarcasm) Think for a minute about the bandwidth of the person(s) w
"So you become the master build engineer. That's a far cry from being a software programmer."
Never claimed I was a software programmer. That wasnt the point I was making. Please re-read and understand my comment.
"Since LFS is primarily intended to teach a person how to set up, configure, and compile existing source code, it is not so much related to programming as it is to painting by numbers."
Your field of vision is rather narrow. You get out what you put in. Nowhere does it say that you cannot use LFS as the start and go further.
LFS is at least a step beyond Gentoo in that you have to do things yourself, not rely on a tool to do it for you.
Now, to reinforce the point I was originally trying to make: Somebody who has a job as a sysadmin, and we're talking 3rd line here - best described as "the last line of defence before (in the case of non-OSS) we call the vendor" - will be well-served by having taken the time to dig into the system; build it, break it, change it, fix it.
What LFS *does* do is provide a leg-up for those people who would otherwise be daunted by said task.
I did the installation-from-sourcecode back in in 1994 using the slackware (2.0) source set, but building manually instead of using the scripts.
At the time I was also a modest C programmmer (2nd year at uni), and took the time to look at the internals of many of the daemons and other major components. It taught me a damn lot.
Eleven years on... I've been a professsional unix sysadmin/engineer for around 8 years, with better understanding of unix - ANY unix, the skills are very transferable - than most others in my organisation with twice my experience, because I have an "internal" view of unix in addition to the "external" view your average admin has.
So the exercise most definitely IS worth the effort.
LFS is NOT about "custom compilation benefits", and is ALL about tracking down those "stupid dependency problems" in order to learn how the whole show hangs together.
One particular advantage of hw raid over sw raid which I do not believe has been mentioned is the removal of complexity.
In particular, given a linux/windows (/anything else) dual boot arrangement, *sharing* a sw raid setup is going to be difficult, because each OS will arrange the raid layout on disk slightly differently.
I have hw raid5 on scsi with a large FAT32 data volume (best common FS between linux/windows), and no problems. I would not like to try to build this using sw raid (and I'd have to pay more for a server edition of windows rather than win2k pro, assuming windows dynamic volumes can *do* raid5).
On the other hand, if the raid board breaks, you're slightly screwed, unless it's recent and/or common enough that you can get a replacement (hint: buy a spare).
Re:The Roman Empire is back - UK and EU
on
BBC to Try TV On Demand
·
· Score: 2, Interesting
>what are the problems with it that make people in Britain so reluctant to join?
The EU needs to be large in order to be effective. However, the enlargement drains the larger economies to benefit of the smaller ones. At this moment, the UK economy is the strongest in Europe (Germany is still struggling with the deadweight of reunification).
Previous attempts at economic glue - viz the exchange rate mechanism - placed enormous strain on the UK and showed just how unbalanced things are.
Another problem to manifest is the ludicrous inflation Ireland experienced (which they appear willing to endure as they really *do* benefit from the EU slush funds they're using to build roads).
The very *last* member of the EU to "cry foul" when things are going seriously wrong is the UK - too much of this stiff upper lip nonsense.
The EU *needs* the economic resources of the UK but very few other members (particularly not the French, who're the ultimate driving force behind the EU) will think twice about enacting legislation which hurts the UK if it benefits themselves.
Increasing numbers of people in the UK are simply brassed off because this is *supposed* to be a two-way street.
>>NULL is often (void *)0, which is *technically* valid on some architectures (OS notwithstanding). It's just so unlikely a result that it's safe to use.
>NO!!!! (void *)0 is not the address of anything (that the C program can use).
To explain to those who can't follow the point:
I wasn't talking about C. Nor Unix. Physical memory addressing starts at zero. Therefore a pointer whose value is zero is a valid pointer (its address exists).
On many (but not all) CPU architectures, address zero is not writeable, but the implementation of memory handling under unix does not require an "invalid" pointer to be zero - simply outside the addressable memory space for the process. Zero is merely convenient (and likely to be protected on the CPU anyway, in the average case).
My *point* was about in-band error values. Given that different values can be valid or invalid for different operations, functions cannot use consistent in-band errors as the poster to whom I replied noted.
To emphasise:
I was observing that value zero is a valid file descriptor, but not (in C) a valid pointer.
Incidentally, with regard to the C faq, I am aware that the compiler spots the string "(void *)0" and uses the "internal representation for the architecture for an invalid pointer".
In the file descriptor world, -1 is invalid, and therefore a good error value.
NULL is often (void *)0, which is *technically* valid on some architectures (OS notwithstanding). It's just so unlikely a result that it's safe to use.
A better approach is this:
int result=some_argument_to_function;
if(function(&result)==0) {/* handle error */ }
Unfortunately, there are too many places where the argument you want to pass would not directly accessible or assignment to it is inconvenient.
Alternatively you could always set and check errno, which should be 0 on no error. This requires and addition line of code to check errors (and programmers, including me, are lazy).
Moving to out-of-band errors without the above requires cartesian products as return values, which are fine in set theory but would probably choke the average programmer.
As plugins to existing applications are so popular these days, I see this issue as an irrelevance.
Both sides are competing using incompatible creeping featurism. Last I looked, Zip applications where supposed to combine and squash files (and that was enough).
What should be done is to separate the operations:
- file browsing (WinRAR's interface trumps both)
- archiving (combining files)
- compression
- encryption
and implement the latter three as functions of the first using plugins (and let the user choose).
Incidentally, Zip's file format (directory last) sucks. It is practically impossible to do the following using zip:
tar Bcf - . | gzip -1c | rsh -n over_there gzip -dc | tar -C/path -Bxvf -
To this end, plugins suggested above should be written as filters where possible.
I have no problem with browser-like interfaces combining other functions, but the Golden Rule still stands: One Tool, One Job.
The fact of the matter is that more is lost in the translation of the DVD; I don't get surround sound and I don't get extras or outtakes.
Personally I watch DVDs for the content, not the extras, which are a marketing gimick. I'm perfectly happy with DivX encodings.
Using your logic, if the CD included recording session outtakes such as when a musician made a mistake, would you buy it?
Some of these ideas are VERY short sighted
on
State Of The Filesystem
·
· Score: 5, Insightful
I wish people with clever ideas to redesign POSIX namespaces would spend ten years in system administration first so they realise what's involved with managing REAL WORKING SYSTEMS.
Some of the ideas might well lead in useful directions, but some (at least as described in the paper) are plain silly. viz:
1) with overlayed mounts:
suppose my home dir is mounted read-write over a read-only system root, and I do not have a "/bin/prog" in my home dir. Consider:
cp/bin/prog/bin/prog
First time, it copies the system/bin/prog into my home fs - Counter-intuitive to the path semantics. If I run this a second time it copies my copy of/bin/prog over itself - Inconsistent.
2) Attributes in the namespace
We have a rather carefully written setuid chown/chgrp/chmod replacement which can be run by users in an "admin" group, and allows devolution of 1st-line support tasks to nominated users. It won't touch files whose uid/gid is 100, so they can only touch non-system files.
If attributes (file uid) is file/..uid and cp is supposed to handle what chown does, the above breaks big-time. We now need a custom cp replacement. Either that or we have to add an ACL for the admin group to every file we want them to manage, which is a great deal of effort, and likely end up inconsistent.
Contrary to the paper, setuid and PARTICULARLY setgid is NOT going to go away in the real world any time soon, as far as files are concerned. Ports less than 1024 are a different matter and I agree with the document.
3) Consider the number of file descriptors involved if/etc/passwd becomes a hierarchy of files. Just logging in one user will involve multiple open()-read()-close() operations. Whilst these might be efficiently implementable at fs-level, it is still very inefficient in user space, or will at least require a dramatic rethink of unix tools.
One is the identity under multiplication just as Zero is the identity under addition (group theory).
As far as the "unique prime factors" explanation goes, since One is the identity, every number must have, in addition to its prime factors, an infinite series of 1s (as the identity, not a prime), as valid factors.
Zero and One are "special cases", because their properties lead to either infinites or contraditions. It makes as much sense to discuss whether One is prime or not as to discuss the actual value resulting from division-by-zero (outside the Aleph sets, that is).
Problem: how does your hardware know an OS binary is authenticated before the OS is loaded? It can't talk to anything else (no network). It must rely on stored information. If that information is stored, it can either be replaced (flashed) or faked. No difference between this "protection mechanism" and DVD zoning, for example. You'll simply "chip" your PC like you "chip" your DVD player, or PlayStation.
This, more than any other protection mechanism, is NOT PRACTICAL.
Secondly, vote with your feet. It is not serious to suggest that you can legislate that a consumer is not free to choose what they buy. So buy something without DRM (if DRM ever appears, which I doubt). Don't believe me? What about those foreign places where they won't dream of DRM because they don't bow down to the RIAA etc? Suddenly the almighty American IT industry is losing money to overseas companies. Dangerous for any government to comtemplate...
I always credited RMS with a case of sour grapes since Linux has, to an extent, stolen a march on the FSF. Admit it, RMS, the Hurd was *floundering*.
But: The GNU Project says, "Please give our project equal mention," but Linus says, "Don't give them a share of the credit; call the whole thing after my name alone!" Now envision the mindset of a person who can look at these events and accuse the GNU Project of egotism.
To the contrary, RMS, your comment quite reveals your own. There is *no way* Linus has that sort of ego. He is not, after all, the one who stamps and screams like a spoiled brat.
Secondly, as regards Bitkeeper. I'm not familiar with either the program or its usage terms, BUT: I do know that to claim that anyone (even those developing OSS) *may not* or *should not* be FREE TO CHOOSE what they use is hypocritical in the extreme coming from a so-called advocate of freedom.
For the record, I prefer plain "Linux" as the name, just like I prefer plain "Solaris", "AIX", etc, where may of the tools derive from ancient BSD/SysV origins. If someone else prefers GNU/Linux, fine (their choice), but I take extreme distaste to anyone who attempts to force their opinion on someone else.
As far as the FSF's "largest single constituent" argument is concerned - rubbish. If I build a car from a kit or spare parts, and *happen* to drop in a Ford engine (as a single largest component) that does *NOT* make it a Ford car. In fact, Ford would take considerable exception if I called it such!
bwoodring & dvc - include & self-fulfulling predictions
What is your view on the ethics of producing results which are used specifically to deceive and manipulate? I appreciate that your business is merely providing a service, but since that service often boils down to "whoring for figures" to support some funders marketing spin (i.e. they
pay you, you give them what they want), can you (as a business) command any respect for what you do, and for your reports?
Slashdot Mirror
A more interesting foil to this comment is "so is it property or is it a service?"
It can't be both. If it's a service, it's not intellectual "property". If it's property, "theft of service" is precluded.
Of course, it's neither, but it's nice to see your opponent squirm on a pointy bit of logic.
I'll second al broccoli's response - the parent post is talking absolute crap.
I, we - the team I work in, have recently finished a migration of SAP4.6 on windows to SAP4.7 on solaris. The windows installation was formerly the largest single installation of SAP on windows in western europe, and it had severe scaling and support problems.
I also know from some years back, this having been discussed in an interview I attended, that Unilever, who before they switched held the crown for the largest installation of SAP on windows, switched to Digital UNIX (before it was Tru64) for exactly the same reason.
SAP on any OS has a hefty hardware requirement list. In addition, to my mind, they make some stupid recommendations about memory. Viz - for our setup we have multiple V440 and V490 suns with 16GB memory, and SAP want (and, on some servers, having hit this problem, *need*) THREE or FOUR TIMES that in swap. I would have suggested either less databases on each machine, or more memory (not sure if 16GB is the physical limit for V440/490) or just bigger machines, but then it's not my job to spec these things.
Anyway, having vast quantities of swap actively used as working memory may have contributed to the instability of a SAP system on linux, if as someone suggested that VM on linux is currently not as good as it could be.
Firstly, thank you for a much more informative reply than your initial complaint.
;)
I still have the impression that you dont have a lot of patience, and that you're so comfortable with BT that anything which doesnt do things in exactly the same way is not good enough. Please remember that neither is good reason to criticise ed2k/emule. Clearly, BT suits your needs much better than ed2k, but that doesnt make BT "right" and ed2k "wrong".
Incidentally, please excuse my aggressive tone where I refute some of your points - it is not intended to offend.
---
I am still not convinced that emule is confusing or difficult to set up. You have not actually identified *what* the problems are. Yes, elitist attitudes are a problem, and yes, it is good that you work to develop autonomic software, but this is just arm-waving in relation to emule, not a proper complaint.
Also, the very popularity of the ed2k network provides empirical evidence that several million people *can* get it working rather well. That BT is *more* simple does not imply emule is hard to set up, only *comparitively* harder.
Servers do come and go on the list. However, as you have not eliminated, nor it seems evaluated the liklihood, that many of those servers are unreliable, point the blame at the emule client is not justified. There are a number of reliable, long-standing servers, and they tend to have large numbers of connected users. A large number of servers appear with just one user, suggesting to me that they are set up "because it's fun". In any case, emule handles variable server presence - you should examine the static server list and automatic reconnection options.
Bootstrapping: emule's kad implementation avoids *all* servers by design; it eliminates the dependence on servers, which is the goal of a fully p2p network.
Azureus' approach may well be more *convenient* to you logging in. "Effective" depends on point of view and stated purpose (such as there is one, of the network).
Server lists: http://www.google.co.uk/search?hl=en&q=ed2k+server +list&btnG=Google+Search&meta=
The one you want is gruk.org, #8 on the list. The #1 entry *does* have some working servers in its list, though the list is old. Many links have server.met files which you'd need to put in the emule config directory - not intuitive, I grant you, though the guide at emule-project.net does describe this, I believe.
With regard to BT's apparent speed over emule, you might want to try the following: set BT to download something large and popular, say, an entire season of Babylon 5, or Galactica. Do the same on ed2k. I grant you, ed2k may take an hour or two to get started, but in my experience it can easily saturate a 2MBit line. I'm sure BT will do the same.
emule is perfectly capable of moving large amounts of data in a short time. However, it may not do it in the order *you* want. If that's not convenient, fine, stick with BT, but *do not* criticise emule for not being the tool it is not designed to be.
(side note: new releases: TV prog shown friday, emule download complete by saturday evening - not as fast as BT, but then I have patience
Searching: yes, ed2k's search is, well let's say erratic. Because it *has* a search, the equivalents of piratebay are not required, but an optional extra (so the quality of such sites are likely to be reduced - though realworld and anidb.info are, as far as I can see, excellent). Realworld is under legal attack, just like Loki and Suprnova were - again, you cannot criticise emule for suffering from the same legal problems as BT.
Useless peers: you can't tell whether a client is *refusing* to upload to you unless you can see its client list (which you can't). All you can see is your position on the queue and yes, sometimes you can be #1 in the queue and drop down without getting anything - caused by re-ask
I'm not going to answer every point, because there's far too much, and it strikes me that you know very little about emule/ed2k and haven't tried very hard to find out.
1) Setup and use is much too confusing.
Confusing? How? Did you examine the guide at emule-project.net? Or are you just assuming that because BitTorrent (BT) does everything for you emule will do the same?
Substantiate and justify this please. (Lists of servers are largely irrelevant. One server with a large number of users is enough. See +++ below.)
2) It doesn't "just work". Getting your client to connect to the kademlia network is a nightmare, ...
Kademlia (kad) is fully p2p, no servers. You need previously-known contacts to connect. Every time you download or upload a file part from/to a client which *is* on kad, your client records that person as a contact. When you start off, you have no contacts, you need to be patient, and in any case you dont have to use kad.
For the server method, you only *need* one server to start - your client will learn about other servers +++ from other clients as you exchange file parts. You can *google* for server lists too, if you must. There may even be a server list linked from emule-project.net
3) Downloads are slow.
Downloads are not *instant*. Yes, there is the queuing system. Please explain to me why you should jump ahead of everyone else who is already waiting for a file? The source's bandwidth is a limited resource. emule slices by time, I'm guessing BT slices by speed. Incidentally, if you are downloading a large, popular file, receiving several parts are once, emule can really eat bandwidth. It comes down to the number of people sharing a file, ultimately.
4) It is hard to search.
At least it *has* a built-in search facility. Different search results are caused by the nature of the search mechanism (more so for kad). The search is *not* an index (unlike, say, google). Also, emule tends not to do a full search if it quickly finds >50 unique matches. And the total number of unique matches is limited to 300. With BT, you're searching (fixed) index sites. emule has equivalents, such as the-realworld, osloskop, osiolek, if you care to look.
As a side note, the very presence of a file in an emule-visible place on your computer means it can be found by someone else through a search, by just the filename. By contrast with BT (I believe) if someone doesnt have a .torrent for a file AND publish it via some *non-BT* method, such as a website, you cant ever get at a file being shared. That is, ed2k provides a way to get the filehash (torrent) without *having* the filehash.
5) eMule "swarms" have tons of useless peers....In a BitTorrent swarm, EVERYBODY is uploading
You cant upload before you have parts of the file *to* upload, true of emule and BT. And with emule you *must* share/upload those parts you already have. The emule credit system promotes (through queues) those who upload (RTFM for details). You *could* modify a client to *not* upload, but it would hurt you, because you'll sit in queues for longer. Incidentally, emule doesnt have "swarms" per se. Overnet did.
Until *you* have parts of a file to upload, *you* are a "useless peer". "Useless peers" cease to be useless when they acquire file parts.
You dont think these two are connected? People *still* sharing things they got >6 months ago while downloading other things? (There is also no "seeding" per se on emule.)
I spent a week downloading a 90MB file,...
(sarcasm) You novice! (/sarcasm) Think for a minute about the bandwidth of the person(s) w
"So you become the master build engineer. That's a far cry from being a software programmer."
Never claimed I was a software programmer. That wasnt the point I was making. Please re-read and understand my comment.
"Since LFS is primarily intended to teach a person how to set up, configure, and compile existing source code, it is not so much related to programming as it is to painting by numbers."
Your field of vision is rather narrow. You get out what you put in. Nowhere does it say that you cannot use LFS as the start and go further.
LFS is at least a step beyond Gentoo in that you have to do things yourself, not rely on a tool to do it for you.
Now, to reinforce the point I was originally trying to make: Somebody who has a job as a sysadmin, and we're talking 3rd line here - best described as "the last line of defence before (in the case of non-OSS) we call the vendor" - will be well-served by having taken the time to dig into the system; build it, break it, change it, fix it.
What LFS *does* do is provide a leg-up for those people who would otherwise be daunted by said task.
You're probably a troll, but I'll play.
"Gotta love how you managed to insinuate that you actually programmed for a living."
I said nothing of the sort. The term I used was engineer, being someone who builds and/or fixes things.
"I guess if you consider shell scripts *guffaw* to be programs, then maybe..."
MUDs actually. dynamic loading, socket programming, command parsing, built-in language interpreter. But just as a hobby.
So back off.
I did the installation-from-sourcecode back in in 1994 using the slackware (2.0) source set, but building manually instead of using the scripts.
At the time I was also a modest C programmmer (2nd year at uni), and took the time to look at the internals of many of the daemons and other major components. It taught me a damn lot.
Eleven years on... I've been a professsional unix sysadmin/engineer for around 8 years, with better understanding of unix - ANY unix, the skills are very transferable - than most others in my organisation with twice my experience, because I have an "internal" view of unix in addition to the "external" view your average admin has.
So the exercise most definitely IS worth the effort.
LFS is NOT about "custom compilation benefits", and is ALL about tracking down those "stupid dependency problems" in order to learn how the whole show hangs together.
One particular advantage of hw raid over sw raid which I do not believe has been mentioned is the removal of complexity.
In particular, given a linux/windows (/anything else) dual boot arrangement, *sharing* a sw raid setup is going to be difficult, because each OS will arrange the raid layout on disk slightly differently.
I have hw raid5 on scsi with a large FAT32 data volume (best common FS between linux/windows), and no problems. I would not like to try to build this using sw raid (and I'd have to pay more for a server edition of windows rather than win2k pro, assuming windows dynamic volumes can *do* raid5).
On the other hand, if the raid board breaks, you're slightly screwed, unless it's recent and/or common enough that you can get a replacement (hint: buy a spare).
>what are the problems with it that make people in Britain so reluctant to join?
The EU needs to be large in order to be effective. However, the enlargement drains the larger economies to benefit of the smaller ones. At this moment, the UK economy is the strongest in Europe (Germany is still struggling with the deadweight of reunification).
Previous attempts at economic glue - viz the exchange rate mechanism - placed enormous strain on the UK and showed just how unbalanced things are.
Another problem to manifest is the ludicrous inflation Ireland experienced (which they appear willing to endure as they really *do* benefit from the EU slush funds they're using to build roads).
The very *last* member of the EU to "cry foul" when things are going seriously wrong is the UK - too much of this stiff upper lip nonsense.
The EU *needs* the economic resources of the UK but very few other members (particularly not the French, who're the ultimate driving force behind the EU) will think twice about enacting legislation which hurts the UK if it benefits themselves.
Increasing numbers of people in the UK are simply brassed off because this is *supposed* to be a two-way street.
>That should be sKeptical. Dammit
Really?
skhool ?
skooter ?
skoop ?
skope ?
skone ?
skallop ?
Get with the beat - English is not consitently phonetically spelled.
(As an aside - my first *guess* would be that sc-spellings are derived from latin whereas sk-spelling are from greek)
Bzzzt. Wrong.
The (academic) definition of "operating system" is "interface to the hardware".
"kernel" in unix terms is synonymous with "operating system".
"distro" is "operating environment".
If you *really* want a complete rant on this, I have written one, sad individual that I am.
In future, please refer to a recognised textbook (suggestions in the link above) before jumping in.
>>NULL is often (void *)0, which is *technically* valid on some architectures (OS notwithstanding). It's just so unlikely a result that it's safe to use.
>NO!!!! (void *)0 is not the address of anything (that the C program can use).
To explain to those who can't follow the point:
I wasn't talking about C. Nor Unix. Physical memory addressing starts at zero. Therefore a pointer whose value is zero is a valid pointer (its address exists).
On many (but not all) CPU architectures, address zero is not writeable, but the implementation of memory handling under unix does not require an "invalid" pointer to be zero - simply outside the addressable memory space for the process. Zero is merely convenient (and likely to be protected on the CPU anyway, in the average case).
My *point* was about in-band error values. Given that different values can be valid or invalid for different operations, functions cannot use consistent in-band errors as the poster to whom I replied noted.
To emphasise:
I was observing that value zero is a valid file descriptor, but not (in C) a valid pointer.
Incidentally, with regard to the C faq, I am aware that the compiler spots the string "(void *)0" and uses the "internal representation for the architecture for an invalid pointer".
There are ways, however, to force the issue:
intptr_t ret_arg(intptr_t arg)
{
return arg;
}
void *ptr=(void *)ret_arg(0);
The ROOT CAUSE here is in-band error values.
/* handle error */ }
In the file descriptor world, -1 is invalid, and therefore a good error value.
NULL is often (void *)0, which is *technically* valid on some architectures (OS notwithstanding). It's just so unlikely a result that it's safe to use.
A better approach is this:
int result=some_argument_to_function;
if(function(&result)==0) {
Unfortunately, there are too many places where the argument you want to pass would not directly accessible or assignment to it is inconvenient.
Alternatively you could always set and check errno, which should be 0 on no error. This requires and addition line of code to check errors (and programmers, including me, are lazy).
Moving to out-of-band errors without the above requires cartesian products as return values, which are fine in set theory but would probably choke the average programmer.
odump -Dl file
.liblist section
reads the
odump -Dx for varities of x reads other interesting parts of the binary header.
Oops. For the script kiddies that should be:
/path -Bxvf -'
tar Bcf - . | gzip -1c | rsh -n over_there 'gzip -dc | tar -C
And YES, I know there are Good Reasons why zip has the directory last. I just don't see they're universally necessary.
As plugins to existing applications are so popular these days, I see this issue as an irrelevance.
/path -Bxvf -
Both sides are competing using incompatible creeping featurism. Last I looked, Zip applications where supposed to combine and squash files (and that was enough).
What should be done is to separate the operations:
- file browsing (WinRAR's interface trumps both)
- archiving (combining files)
- compression
- encryption
and implement the latter three as functions of the first using plugins (and let the user choose).
Incidentally, Zip's file format (directory last) sucks. It is practically impossible to do the following using zip:
tar Bcf - . | gzip -1c | rsh -n over_there gzip -dc | tar -C
To this end, plugins suggested above should be written as filters where possible.
I have no problem with browser-like interfaces combining other functions, but the Golden Rule still stands: One Tool, One Job.
The fact of the matter is that more is lost in the translation of the DVD; I don't get surround sound and I don't get extras or outtakes.
Personally I watch DVDs for the content, not the extras, which are a marketing gimick. I'm perfectly happy with DivX encodings.
Using your logic, if the CD included recording session outtakes such as when a musician made a mistake, would you buy it?
I wish people with clever ideas to redesign POSIX namespaces would spend ten years in system administration first so they realise what's involved with managing REAL WORKING SYSTEMS.
/bin/prog /bin/prog
/bin/prog into my home fs - Counter-intuitive to the path semantics. If I run this a second time it copies my copy of /bin/prog over itself - Inconsistent.
/etc/passwd becomes a hierarchy of files. Just logging in one user will involve multiple open()-read()-close() operations. Whilst these might be efficiently implementable at fs-level, it is still very inefficient in user space, or will at least require a dramatic rethink of unix tools.
Some of the ideas might well lead in useful directions, but some (at least as described in the paper) are plain silly. viz:
1) with overlayed mounts:
suppose my home dir is mounted read-write over a read-only system root, and I do not have a "/bin/prog" in my home dir. Consider:
cp
First time, it copies the system
2) Attributes in the namespace
We have a rather carefully written setuid chown/chgrp/chmod replacement which can be run by users in an "admin" group, and allows devolution of 1st-line support tasks to nominated users. It won't touch files whose uid/gid is 100, so they can only touch non-system files.
If attributes (file uid) is file/..uid and cp is supposed to handle what chown does, the above breaks big-time. We now need a custom cp replacement. Either that or we have to add an ACL for the admin group to every file we want them to manage, which is a great deal of effort, and likely end up inconsistent.
Contrary to the paper, setuid and PARTICULARLY setgid is NOT going to go away in the real world any time soon, as far as files are concerned. Ports less than 1024 are a different matter and I agree with the document.
3) Consider the number of file descriptors involved if
"Two billion years ago, in a far-away galaxy, a giant star exploded."
The death of star. Death Star.
I predict they might be seeing a second one of these explosions any time soon...
One is the identity under multiplication just as Zero is the identity under addition (group theory).
As far as the "unique prime factors" explanation goes, since One is the identity, every number must have, in addition to its prime factors, an infinite series of 1s (as the identity, not a prime), as valid factors.
Zero and One are "special cases", because their properties lead to either infinites or contraditions. It makes as much sense to discuss whether One is prime or not as to discuss the actual value resulting from division-by-zero (outside the Aleph sets, that is).
This, more than any other protection mechanism, is NOT PRACTICAL.
Secondly, vote with your feet. It is not serious to suggest that you can legislate that a consumer is not free to choose what they buy. So buy something without DRM (if DRM ever appears, which I doubt). Don't believe me? What about those foreign places where they won't dream of DRM because they don't bow down to the RIAA etc? Suddenly the almighty American IT industry is losing money to overseas companies. Dangerous for any government to comtemplate...
But: The GNU Project says, "Please give our project equal mention," but Linus says, "Don't give them a share of the credit; call the whole thing after my name alone!" Now envision the mindset of a person who can look at these events and accuse the GNU Project of egotism.
To the contrary, RMS, your comment quite reveals your own. There is *no way* Linus has that sort of ego. He is not, after all, the one who stamps and screams like a spoiled brat.
Secondly, as regards Bitkeeper. I'm not familiar with either the program or its usage terms, BUT: I do know that to claim that anyone (even those developing OSS) *may not* or *should not* be FREE TO CHOOSE what they use is hypocritical in the extreme coming from a so-called advocate of freedom.
For the record, I prefer plain "Linux" as the name, just like I prefer plain "Solaris", "AIX", etc, where may of the tools derive from ancient BSD/SysV origins. If someone else prefers GNU/Linux, fine (their choice), but I take extreme distaste to anyone who attempts to force their opinion on someone else.
As far as the FSF's "largest single constituent" argument is concerned - rubbish. If I build a car from a kit or spare parts, and *happen* to drop in a Ford engine (as a single largest component) that does *NOT* make it a Ford car. In fact, Ford would take considerable exception if I called it such!
Begone, Bigot!
- connordb - results for funder not the public
- dmccarty - palm OS predictions wrong
- bwoodring & dvc - include & self-fulfulling predictions
What is your view on the ethics of producing results which are used specifically to deceive and manipulate? I appreciate that your business is merely providing a service, but since that service often boils down to "whoring for figures" to support some funders marketing spin (i.e. they pay you, you give them what they want), can you (as a business) command any respect for what you do, and for your reports?