Apparnely, the biggest CO in town has 4' raised floors, and somewhere down there there is a massive stuffed owl. "Hey, new guy! Get down there and fix pair XYZ!"
The Gimp people seem like they have gone out of their way to do things differently then Photoshop. But the lack of a master window is one thing that paticularly pisses me off too.
It is worth pointing out that the quid pro quo behind Copyright and Patents are the same. One applies to technology, one to art, but the "deal" is the same: you (a creator) create something new and the government (scocity) will offer you protection for it, for a limited period of time.
The quid pro quo with trademarks is a little different. It boils down to confusion. It is in everyones best interest that names are unique within a feild. Some people might actualy think they are getting a high quality watch when they buy a Rollex on the street. Trademarks protect both the consumer (from knock offs (amongst other things)) and the producers
(from lost sales, cost marketing others products, etc).. But if the producer cant take the trouble to protect their trademarks, then scocity (government) isnt going to help them.
The question is: Did you write a benchmark program that was desigined in a hardware-neutral way? Or did you write a program to do X, highly optimized to do it the Sun/UltraSparc/Solaris way, but also managed to get it to run on a lintel box?
SATA, by definition, is hot swappable. Perhaps your case doesnt handle it, but its part of the spec.
This may be a compleate non-issue for a "workstation".
Which is kinda like SCSI/SCA. It is also, by definition, hot swappable. Provided your case supports it. Which, for example, the Ultra 1 did not.
On the other hand, it is only a matter of time before PATA drives start becoming very hard to find. One of the theoretical advantages of Sun hardware is that it has very long support. In 5 years time if you call up Sun for a SunApproved (tm) PATA drive for your Blade 1500, one of three things will happen: a) they will get one that has been sitting on the shelf for 5 years from their big, expensive, warehouse; b) they will get one from a custom manufacturing run; or c) they will say you are SOL. Being Sun, c) is unlikely. But the other 2 options will cost you around $500 for a (at that point) $3.50 drive.
That Lucas was heavily influenced by The Hidden Fortress.. It is the story of to peasants/slaves. If you note, the only characters who have been in all of the Star Wars films so far are the two droids.
It is the droids who pull the whole story together. This is a intentional, if loose goal.
Sort of like how what could have been called "A breif history of US events from 1945-1980 as witnessed by a baffoon" ended up being called "Forest Gump"
I have a DSL connection at home, and I work from home. My router/firewall/server and desktop are always turned on. I always have a copy of mozilla running, and evolution is up most of the time.
Does that count as being online 168 hours a week?
If Im working on some programming project for, say a 4 hour streach, and Im flipping back and forth to a browser pointed at some online documentation, does that count as 4 hours online? Or (pulling a number out of my ass) is only 10% of that online?
When were you a kid? MAD magazine was printed on about the most cheap paper avialable, and only recently changed.
Once, the printer of MAD called up the offices to tell William Gaines (the founder, publisher) that he was unable to get the cheap paper for a given months print run, but not to worry, he would suck up the difference. Gaines refused and demanded that the cheap stock be used. "Our readers expect MAD to be printed on cheap stock! Dammit, I demand cheap paper!"
The best tool for the job is the tool that has already been purchased in most cases. Powerpoint comes with all MS Office versions since 97 (At least). And its not all that bad.
It matters not that (the topics program) is free, or that OpenOffice is free, or that...... What matters is that Powerpoint is alreday been paid for, installed, and people have a bit of a clue of how to use it (since it uses common controls with the rest of Office).
You would have to be doing one fancy ass presentation to warrent spending time/money/effort on something besides Powerpoint if you already have it installed. And chances are that time/effort/money would be on a 3rd party plugin for Powerpoint, rather then a replacement for it.
Actually, the archival quality of film has been generaly discounted.
Nitrate film (used almost exclusivly untill 1950) is a legendary fire safety hazard, even if specific accidents have been rare. All film degrades, even if stored properly. And proper storage of "important" film hasent always happened, nor is it happening now.
The only guarenteed archival method is to digitize (whatever) at a higher resolution then the origional, stored uncompressed, or at least with a non-lossy compression method. Document the storage method, And then every 10 years or so move it to a new digital medium, and if necessary the new format.
We are in a state today that we have lots of digital data that is all but useless. For many things we have neither the physcial devices to read them. And for things that we can read the bits, we dont have the documentation for what the format is. No one make 9 track tape drives anymore, for example... A group at NASA maintains even older drives, by hand, assumably at enormous cost.
Bayesian filters wont catch gibberish, they will catch specific giberish. As will the rule based ones. (or not depending on how good the rules and/or training is).
Re:Sends binary files as text/plain MIME type
on
2003: Year of Apache
·
· Score: 1
I did read your links, and I agree that Apache is doing the wrong thing.
But your missing my point. Apache is not sending you a file; a given website is sending you a file. The website is confusing you browser. The website can fix the problem by fixing their mime definitions. The website has chosen to not configure their system correctly, and to use software that does the wrong thing when its not configured correctly. The website can even complain to the Apache people, and in general about Apache. But you cant.
Further with the hotel analogy: You dont have a relationship with NEC. NEC is not over charging you. NEC is not making mistakes on your bill. The hotel has a responsibility not to overcharge you. If the happen to overcharge people a lot (more then can be accounted for by mistakes) then they should see what is wrong. It could be their procedures, poor training, stupid staff, or faulty equipment. But why your being overcharged is none of your business. And the hotel using faulty equipment is not an excuse.
For number crunching apps, sure. You diddnt mention the use of computers as far back as WWII for number crunching for nuclear research. In WWII it was to figure out how much stuff they needed, now entire nuclear explosions can be simulated. Or so they claim.
But there is a sepearate, distinct, and very important component of computers that isnt realy computation: data processing.
This is where, historicly (<1975 say) where IBM (and its predecessors) worked almost exclusivly. Censuses started it all, the 1890 US Census being the first done on punch card machines (reducing a 10 year job into months). IIRC, czatist Russia leased Hollerith machines in the 19th century. (since censuses doers were the primary market for infintile IBM, and no one continiously took censuses, IBM generaly leased machines (and opearators, assumably) rather then selling them. Of course, they continue this practice, esp. on the "big iron", even today).
Most people agree that censuses, at least, are benign. It hasent been until the last 5 years that data processing has become sufficently advanced for average people to consiter it at all threatining. Im making a distinction beteween data collection/processing itself from the application therof. Privacy concerns (for example) are now very much a concern of "normal people", even if they otherwise trust the data collectors and what happens to the data. Up untill 5 years ago no one had enough data for the data alone to be risky/dangerous/intrusive. Now, not so much. Anyway...
On the other hand, "data processing", even before "computation", has been used for what would be universally accepted as evil purposes. Or at least one: I speak of Nazi Germany using Hollerith machines to keep tabs on the Jews. To quantify the "Jew problem" (as they saw it). And to effectivly round them up. The rest being "common" history (which I will ignore, this being a discussion of computers). The use of Hollerith machines being largely unknown, even amongst computer/IT types. Even though I dont agree with the authors basic premis (that IBM is at least morally liable for some of the Holocaust), I will point out IBM and the Holocaust : The Strategic Alliance Between Nazi Germany and America's Most PowerfulCorporation.
Of course the topic of discussion here is military usage of computers. Censuses certenly dont count. I dont think the Nazis use of computers does either. There is a distinction beteween the German Military/Navy, and the German (Nazi) Government, and "special" (ie, SS) forces.
Re:Sends binary files as text/plain MIME type
on
2003: Year of Apache
·
· Score: 1
Ok. You have a relationship with a website. The sysadmins of the website have a relationship with Apache. You are fully within your rights to bitch and complain about your favourite website. You have exactly 0 rights to complain about Apache. If you, as a user, have a problem whose root cause is the fault of Apache (which, in this case, is only half true), your problem is with your website. Yes, apache may be doing the wrong thing. But the server admins can fix it. Easily. One line per file type. 20 seconds.
When will users learn to complain to the right people? Users have relationships with ISPs, content providers, hosting providers, whatever. The server admin guys have relationships with the server-software developement group.
If you go to a hotel, do you complain to Serta that your bed is lumpy? Hell no. You compain to the hotel. The problem may be that a given matteress sucks, and the mattrss company might be at fault. But did you choose a mattress? No, you chose a hotel, and its the hotels job to provide a good service.
In the last month Ive seen an end user complain about Horde/IMP on dev@horde.org. WTF? If your not a server admin you shouldnt be bothering anyone at horde.org, lel alone dev@. Ive also seen someone complain on the Mailman dev list that a spammer is using Mailman to send out spam, and not honoring unsubscribe requests. As if this is Mailmans fault and the Mailman people should put in a "trojan" style fix so when the spammer upgrades this user can unsubscribe.
Re:Sends binary files as text/plain MIME type
on
2003: Year of Apache
·
· Score: 1
Thats not Apaches fault. Its your fault. Fix your httpd.conf mod_mime docs
The Apache people has a clear policy on mime types, described at the bottom of the linked page:
Please do not send requests to the Apache HTTP Server Project to add any new entries in the distributed mime.types file unless (1) they are already registered with IANA, and (2) they use widely accepted, non-conflicting filename extensions across platforms. category/x-subtype requests will be automatically rejected, as will any new two-letter extensions as they will likely conflict later with the already crowded language and character set namespace.
Lies, damm lies and statistics.
on
2003: Year of Apache
·
· Score: 2, Interesting
While these numbers are impressive, and definitly show relative growth of a given product over time, there usefellness compating products is suspect.
If raw counts of usages indicate quality, then MSIE would be the highest quality web browser by a factor of around 20 (something 95% market share right?). Outlook would be the best mail/PIM software./. readers would disagree with such a statement. So why do we accept conclusions based the same type of logic based on stats from netcraft?
What "we" need is something like some the stock market indicators. [the good ones] are not just a raw sum of all the stocks out there, or all the stocks traded on a given market. There a collection of hand picked stocks. I suspect the specific criteria for being included are secret, but long term stability is almost definitly an important peice of the pie. There not using penny stocks, just IPOd companies, companies in trouble, or companies experiencing isolated/unique growth.
What I propose for someone to do, is to develop such a system for HTTP server usage. Build a list of say, 5000, sites. The sites should be distributed accross all topics, all markets. It should include sites run by non-IT centric companies, IT companies that are primarly "brick and mortar" and web-only companies as well. It should include scanning web hosting companies, colo housed sites, sites run off 56k modems. What they have in common is that they all have some level of longevity (if not stability).
=~ isnt a function. Its an opearator. Perl has string manipulation facilities buit into it at a very low level. PHP does not.
PHP may be clearer, but that is because you (obviously) dont know what the =~ opearator does. Which, if you've never seen Perl code before, is quite resonable. Lets say you dont know what + does:
Assuming you dont know what + does, the first example is the most clear. The last example is the least clear, but I suspect is the most common way to add 2 to something. Since addition is a very common thing to do, most languages have + built in as a native opearator. Perl is desigined to do string manipulation, so it has string opearators built into the language.
If you using grep then your not using PHP. $usefull = system("grep $file"); does not count as using PHP. If you want to use the UNIX tool set, then use the UNIX tool set.
PHP is not paticularly good at manipulating text. It is infinitly better then C, or Modula. But compared to gawk, sed... the unix toolkit, let alone Perl, it sucks. The PHP developers know this, and they have wraped up some of espically usefull features of Perl in to PHP functions... Things that Perl treats as natiave opearators. Compare equivelent search/replace code.
PHP:
$data = preg_replace("/hello/", "goodby", $data);
Perl:
$data =~ s/hello/goodby/;
Minimizing things down to 1 character variable names, 1 character strings/regular expressions, PHP has a character count of 35. Perl is less then half at 14. Because Perl has string manipluation facilities built into the core language. String manipulation is Perl's first, and primary design goal.
I use PHP all the time. And Im probabaly a better PHP programmer then a Perl programmer. But for text analysis, Perl is king.
First of all; he already has logs. He asked about log analysis, not log creation. If you mean to insert a custom php function into every file dologline() or something, that would be a) slow, b) hard-to-impossible to implement on potentially thousands of files c) incompatable if he is using any other server generated pages. And it would miss static objects, images for example.
If you mean, write your own log parser in PHP, again that would be both slow and painfull. PHP isnt paticularly good at text manipulation (which is ironic, seeing as how thats what html generation basicly is). And even if it was as good (or better then), say Perl, Perl has massive amounts of free modules avialable desigined for log analysis, statistics, report generation. PHP is lacking in the third party lib field.
I dont know the state of Activestate (*cough*) so I dont know how well existing, built for Unix, Perl log analysis packages would work. I suspect that many just would. To which, I point out a freshmeat link.
Slashdot Mirror
The entire high school has 16mb of RAM? For fuck sake man, upgrade to a PDP-11!
hehe
punch down block
William! Shatner! Is! A! Spokesman! For! Priceline! Dot! Com! Not! Yahoo!
The Gimp people seem like they have gone out of their way to do things differently then Photoshop. But the lack of a master window is one thing that paticularly pisses me off too.
The quid pro quo with trademarks is a little different. It boils down to confusion. It is in everyones best interest that names are unique within a feild. Some people might actualy think they are getting a high quality watch when they buy a Rollex on the street. Trademarks protect both the consumer (from knock offs (amongst other things)) and the producers (from lost sales, cost marketing others products, etc).. But if the producer cant take the trouble to protect their trademarks, then scocity (government) isnt going to help them.
The question is: Did you write a benchmark program that was desigined in a hardware-neutral way? Or did you write a program to do X, highly optimized to do it the Sun/UltraSparc/Solaris way, but also managed to get it to run on a lintel box?
This may be a compleate non-issue for a "workstation".
Which is kinda like SCSI/SCA. It is also, by definition, hot swappable. Provided your case supports it. Which, for example, the Ultra 1 did not.
On the other hand, it is only a matter of time before PATA drives start becoming very hard to find. One of the theoretical advantages of Sun hardware is that it has very long support. In 5 years time if you call up Sun for a SunApproved (tm) PATA drive for your Blade 1500, one of three things will happen: a) they will get one that has been sitting on the shelf for 5 years from their big, expensive, warehouse; b) they will get one from a custom manufacturing run; or c) they will say you are SOL. Being Sun, c) is unlikely. But the other 2 options will cost you around $500 for a (at that point) $3.50 drive.
Thats easy. More SCO stories and dups.
It is the droids who pull the whole story together. This is a intentional, if loose goal.
Sort of like how what could have been called "A breif history of US events from 1945-1980 as witnessed by a baffoon" ended up being called "Forest Gump"
Does that count as being online 168 hours a week?
If Im working on some programming project for, say a 4 hour streach, and Im flipping back and forth to a browser pointed at some online documentation, does that count as 4 hours online? Or (pulling a number out of my ass) is only 10% of that online?
Once, the printer of MAD called up the offices to tell William Gaines (the founder, publisher) that he was unable to get the cheap paper for a given months print run, but not to worry, he would suck up the difference. Gaines refused and demanded that the cheap stock be used. "Our readers expect MAD to be printed on cheap stock! Dammit, I demand cheap paper!"
Sounds like money well spent to me.
It matters not that (the topics program) is free, or that OpenOffice is free, or that...... What matters is that Powerpoint is alreday been paid for, installed, and people have a bit of a clue of how to use it (since it uses common controls with the rest of Office).
You would have to be doing one fancy ass presentation to warrent spending time/money/effort on something besides Powerpoint if you already have it installed. And chances are that time/effort/money would be on a 3rd party plugin for Powerpoint, rather then a replacement for it.
Nitrate film (used almost exclusivly untill 1950) is a legendary fire safety hazard, even if specific accidents have been rare. All film degrades, even if stored properly. And proper storage of "important" film hasent always happened, nor is it happening now.
The only guarenteed archival method is to digitize (whatever) at a higher resolution then the origional, stored uncompressed, or at least with a non-lossy compression method. Document the storage method, And then every 10 years or so move it to a new digital medium, and if necessary the new format.
We are in a state today that we have lots of digital data that is all but useless. For many things we have neither the physcial devices to read them. And for things that we can read the bits, we dont have the documentation for what the format is. No one make 9 track tape drives anymore, for example... A group at NASA maintains even older drives, by hand, assumably at enormous cost.
At some sufficently advanced time in the future when only art types are using film, the film they will be using wont be Kodak.
Bayesian filters wont catch gibberish, they will catch specific giberish. As will the rule based ones. (or not depending on how good the rules and/or training is).
But your missing my point. Apache is not sending you a file; a given website is sending you a file. The website is confusing you browser. The website can fix the problem by fixing their mime definitions. The website has chosen to not configure their system correctly, and to use software that does the wrong thing when its not configured correctly. The website can even complain to the Apache people, and in general about Apache. But you cant.
Further with the hotel analogy: You dont have a relationship with NEC. NEC is not over charging you. NEC is not making mistakes on your bill. The hotel has a responsibility not to overcharge you. If the happen to overcharge people a lot (more then can be accounted for by mistakes) then they should see what is wrong. It could be their procedures, poor training, stupid staff, or faulty equipment. But why your being overcharged is none of your business. And the hotel using faulty equipment is not an excuse.
But there is a sepearate, distinct, and very important component of computers that isnt realy computation: data processing.
This is where, historicly (<1975 say) where IBM (and its predecessors) worked almost exclusivly. Censuses started it all, the 1890 US Census being the first done on punch card machines (reducing a 10 year job into months). IIRC, czatist Russia leased Hollerith machines in the 19th century. (since censuses doers were the primary market for infintile IBM, and no one continiously took censuses, IBM generaly leased machines (and opearators, assumably) rather then selling them. Of course, they continue this practice, esp. on the "big iron", even today).
Most people agree that censuses, at least, are benign. It hasent been until the last 5 years that data processing has become sufficently advanced for average people to consiter it at all threatining. Im making a distinction beteween data collection/processing itself from the application therof. Privacy concerns (for example) are now very much a concern of "normal people", even if they otherwise trust the data collectors and what happens to the data. Up untill 5 years ago no one had enough data for the data alone to be risky/dangerous/intrusive. Now, not so much. Anyway...
On the other hand, "data processing", even before "computation", has been used for what would be universally accepted as evil purposes. Or at least one: I speak of Nazi Germany using Hollerith machines to keep tabs on the Jews. To quantify the "Jew problem" (as they saw it). And to effectivly round them up. The rest being "common" history (which I will ignore, this being a discussion of computers). The use of Hollerith machines being largely unknown, even amongst computer/IT types. Even though I dont agree with the authors basic premis (that IBM is at least morally liable for some of the Holocaust), I will point out IBM and the Holocaust : The Strategic Alliance Between Nazi Germany and America's Most PowerfulCorporation.
Of course the topic of discussion here is military usage of computers. Censuses certenly dont count. I dont think the Nazis use of computers does either. There is a distinction beteween the German Military/Navy, and the German (Nazi) Government, and "special" (ie, SS) forces.
When will users learn to complain to the right people? Users have relationships with ISPs, content providers, hosting providers, whatever. The server admin guys have relationships with the server-software developement group.
If you go to a hotel, do you complain to Serta that your bed is lumpy? Hell no. You compain to the hotel. The problem may be that a given matteress sucks, and the mattrss company might be at fault. But did you choose a mattress? No, you chose a hotel, and its the hotels job to provide a good service.
In the last month Ive seen an end user complain about Horde/IMP on dev@horde.org. WTF? If your not a server admin you shouldnt be bothering anyone at horde.org, lel alone dev@. Ive also seen someone complain on the Mailman dev list that a spammer is using Mailman to send out spam, and not honoring unsubscribe requests. As if this is Mailmans fault and the Mailman people should put in a "trojan" style fix so when the spammer upgrades this user can unsubscribe.
The Apache people has a clear policy on mime types, described at the bottom of the linked page:
Please do not send requests to the Apache HTTP Server Project to add any new entries in the distributed mime.types file unless (1) they are already registered with IANA, and (2) they use widely accepted, non-conflicting filename extensions across platforms. category/x-subtype requests will be automatically rejected, as will any new two-letter extensions as they will likely conflict later with the already crowded language and character set namespace.
If raw counts of usages indicate quality, then MSIE would be the highest quality web browser by a factor of around 20 (something 95% market share right?). Outlook would be the best mail/PIM software. /. readers would disagree with such a statement. So why do we accept conclusions based the same type of logic based on stats from netcraft?
What "we" need is something like some the stock market indicators. [the good ones] are not just a raw sum of all the stocks out there, or all the stocks traded on a given market. There a collection of hand picked stocks. I suspect the specific criteria for being included are secret, but long term stability is almost definitly an important peice of the pie. There not using penny stocks, just IPOd companies, companies in trouble, or companies experiencing isolated/unique growth.
What I propose for someone to do, is to develop such a system for HTTP server usage. Build a list of say, 5000, sites. The sites should be distributed accross all topics, all markets. It should include sites run by non-IT centric companies, IT companies that are primarly "brick and mortar" and web-only companies as well. It should include scanning web hosting companies, colo housed sites, sites run off 56k modems. What they have in common is that they all have some level of longevity (if not stability).
PHP may be clearer, but that is because you (obviously) dont know what the =~ opearator does. Which, if you've never seen Perl code before, is quite resonable. Lets say you dont know what + does:
Stupid example language:
$i = add_values($i, 2);
PHP (or Perl, C, Java...):
$i = $i + 2;
or:
$i += 2;
Assuming you dont know what + does, the first example is the most clear. The last example is the least clear, but I suspect is the most common way to add 2 to something. Since addition is a very common thing to do, most languages have + built in as a native opearator. Perl is desigined to do string manipulation, so it has string opearators built into the language.
PHP is not paticularly good at manipulating text. It is infinitly better then C, or Modula. But compared to gawk, sed... the unix toolkit, let alone Perl, it sucks. The PHP developers know this, and they have wraped up some of espically usefull features of Perl in to PHP functions... Things that Perl treats as natiave opearators. Compare equivelent search/replace code.
PHP:
$data = preg_replace("/hello/", "goodby", $data);
Perl:
$data =~ s/hello/goodby/;
Minimizing things down to 1 character variable names, 1 character strings/regular expressions, PHP has a character count of 35. Perl is less then half at 14. Because Perl has string manipluation facilities built into the core language. String manipulation is Perl's first, and primary design goal.
I use PHP all the time. And Im probabaly a better PHP programmer then a Perl programmer. But for text analysis, Perl is king.
First of all; he already has logs. He asked about log analysis, not log creation. If you mean to insert a custom php function into every file dologline() or something, that would be a) slow, b) hard-to-impossible to implement on potentially thousands of files c) incompatable if he is using any other server generated pages. And it would miss static objects, images for example.
If you mean, write your own log parser in PHP, again that would be both slow and painfull. PHP isnt paticularly good at text manipulation (which is ironic, seeing as how thats what html generation basicly is). And even if it was as good (or better then), say Perl, Perl has massive amounts of free modules avialable desigined for log analysis, statistics, report generation. PHP is lacking in the third party lib field.
I dont know the state of Activestate (*cough*) so I dont know how well existing, built for Unix, Perl log analysis packages would work. I suspect that many just would. To which, I point out a freshmeat link.