And the answer to that is that some people still use DOS. To some people, DOS is business-critical. If bios makers drop the "cruft", they will lose sales to these people, wheras they gain little (nothing?) from dropping it.
Maintenance of this cruft certainly have a cost. There might even be economy in using smaller Flash chip for the BIOS.
Take the Linksys WRT54G, it's Linux-based. Linksys gains from using the well tested Linux core, and the customers gain by having the option to hack it at will. Check out http://openwrt.org/ [openwrt.org] for an example of the positive results.
The WRT54G firmware is not a BIOS, it is more like the complete OS of an embedded system. Below this firmware is another layer of software (I can't recall the name) whose job is to boot the firmware via tftp if boot_wait is set, for example.
Really, I would not want my BIOS to include a shell and a web server !
The problem with the Maximalist approach as you define it is that you will need to update the firmware every time you install a new piece of hardware. I can't believe my firmware will support every sound card and TV tuner under the sun, for example.
I am not an engineer, but I would go with a Minimalist approach augmented with just enough smart to abstract access to the boot device. Which is, in itself, pretty big if you take into account booting from the network, USB/Firewire devices, etc. From there on, the bootloader or the OS kernel should take control.
Some onboard boot code is certainly required, but I would like the BIOS to stay as minimal as possible. Basically, just enough to load the bootloader.
There is trend toward larger and more featureful BIOS. I just don't understand that. Everything that can be done by the OS should be. Updating an OS is much easier than a BIOS, for starter.
And while you are there, why not drop the DOS-supporting cruft ? (like most of the BIOS software interrupts, for example)
Except that you can't use encryption on your link. Bye bye WEP, WPA, ssh, SSL, PGP etc. I am not even sure you could use simple challenge-based authentication scheme such as APOP. Between broadcasting my passwords for anybody to fetch with 20$ worth of equipement or restricting myself to 4 watts EIRP, the choice is an easy one.
73 de VA2***
(sorry for the incomplete callsign, I very much value my privacy on this board)
Finally, business is the lubrication that prevents the gears of democracy from locking up.
No. Business is about making money. The "lubrication that prevents the gears of democracy from locking up" is citizen implication and conscientisation. Business does pretty well these day while the current state of Western democracy is damn pathetic. Connect the dots.
A lot of free-market idealists complain about a public administration using taxes to fund a service competing with established business. I personnally won't shed a tear over poor telcos losing a little piece of their near monpo^W^W market, but I can see the point.
What would you think, then, of a community network providing free wireless Internet, ran by volunteers and funded with contributions ? Would that still be unfair competition ?
Otherwise it would unfair competition with private companies.
I do not care about the fate of ISP. Network connectivity is fast becoming a commodity, so most are doomed anyway (only the largest ones will survive in the mid-term). What I do care about is ubiquitous and cheap (if not free) access to the Internet. If municipal WISP can provide that in a cost-effective fashion, why not ?
Due to the simplicity of the circuit design, the prototype unit does not discriminate between other sources of 2.4GHz RF, eg. "leaky" microwave ovens, cordless phones, etc.
Considering WiFi networks are just one type of device running on 2.4 Ghz, expect many false positive. In a urban setting, the ISM band is pretty much saturated so your LED will most probably stay a solid green.
Also, I have doubts about the sensitivity of their antenna design.
Most people are only concerned that the product they use functions as it was intended.
Keyword here being "most". I am not "most" people, I am a hacker and a thinkerer. The rest of the world can be happy with black box that "just work", and that is fine by me. But the mere knowledge that I hold in my hand a microprocessor-based device that I can't reprogram nag me in some very annoying way. YMMV. Mere mortal can't understand.
We can see that the hacker spirit have definitely left this board when most answers to such a question is "Why would you want to do that anyway ?".
Claiming the bounceback of IT based on the success of a few gadgets and the stock price of Google seem a bit exagerated. What about spending in network infrastructure and large-scale ERP deployement ? That's what IT is as far as I am concerned.
(last time it was an upgrade of PHP4 on a Red Hat 7.3 I "inherited", that broke a webmail tool)
I bet you talk about the change in the c-client library that broke PHP's imap_open() with the/nottls option. You gotta admit that this is a long time ago, and not even on Fedora either.
Regarding Debian, you are right. But I can't help to find it pathetic that the default kernel of the stable distribution, shipped in 2004, still use kernel 2.2. And you can get the same level of version stability by going with a RHEL clone such as CentOS.
I have witnessed a lot of distro snobbism coming from the Debian crowd lately, especially toward Fedora. Debian might be the best thing since sliced-bread, but it also have a lot of stuff hindering its adoption. The Debian development and release process is pretty counter-intuitive to the laymen, and the stable version is horribly outdated. Actually using Debian also require quite a bit of Debian-specific knowledge (ie I never heard about cron-apt and list-bug before), the default install pretty much suck (no less, elvis for vi, etc), the installer is reminding of 1997, etc. Nothing show-stopping by itself here, but a lot of small hinderance. For beginner, the path of least resistance is often one of the RedHat-derived distro, and Fedora is the best known of these.
With Debian, you will end up using either a distribution based on kernel 2.2, or one labelled "testing". You and I know that Debian testing is pretty solid, and get fix diligently when it break. But so does Fedora.
This brings true desktop search to Windows (for those who don't have Google Desktop Search or similar software running already) and also includes features like
search term highlighting in web pages, auto-completing of forms, and a pop-up blocker.
More power to you if you like Eudora. I don't. We did a very large scale mail server installation (85K accounts) in an organisation that used to have Eudora as the standard MUA. A lot of users of Eudora 5 in the wild. Like everything, bug happen and get corrected so I should not judge the actual quality of a software by it's older releases, but some of them are just too good to pass.
For example, we had to disable STARTTLS IMAP extension because an older release of Eudora was sending STARTSSL. Yes, you read that correctly : STARTSSL. Can you believe that ? That such an obvious bug got passed QA and release astound me. The guy who wrote the STARTTLS support obviously never even bothered to *test* his code.
Another huge source of fuck-up related to Eudora was the habit it had of sending raw 8 bits character in header. In case you don't know, that's a pretty big violation of RFC822. This is particularly thorny as there is no way to tell which charset was used, so automated conversion was not really possible. Did they fixed this one at last ?
There are more I can't remember on the top of my head. Eudora have won the #1 spot in my list of broken MUA thanks to that, above Outlook and OE (who have some interesting way to fuck up too).
Lot's of people in this thread are saying that chead flatbed scanner are really bad for scanning slides. I have a pile of a few hundreds 5x7 family pictures I would like to scan eventually. Budget is very limited, and Linux support is a MUST (no Windows here). I don't necessarily need pro-level or prepress quality, but the nicer the better. I thought about either a Canon CanoLide 20 or 30, as they are cheap and well supported in Linux. How well should they fare ? Any other suggestions ?
And as for reinventing the wheel, it's not really. Samba4 has a generic backend and an ASN1 library. Adding LDAP server-support doesn't require that much. There are some difficult parts but basic functionality isn't so bad.
ASN.1 is a biggy, but it does'nt stop there. Filter parsing, ACI/ACL, the storage mechanism plus all of the network protocol itself. Hopefully, you will just fork OpenLDAP instead of writing from scratch. Samba is a nice project, but it still require a lot of work. I'd rather see you guys woking on QA and solving exisiting problems.
If Samba4 does get it's own LDAP server, you may seem some extremely good interop in Samba4.
This strike me as being a good example of wheel-reinventing and NIH syndrome. And reinventing a pretty complex wheel at that; LDAP is lightweight only in name, that's for sure. Would'nt it be easier to add the required functionnality and behavior to OpenLDAP, maybe as a plugin/module, a build option or as a runtime configuration directive ?
LDAP is used to centralize information. As a Samba user, I would prefer Samba to use my existing information store if at all possible. Being forced to build a separate DIT just for my Samba stuff is not exactly a rejoicing thought and imply a lot of duplication I am trying to avoid in the first place by using LDAP.
Any pointer toward good discussion about why existing LDAP daemon can't be used for AD interop ? I guess ACI support on individual object is a big one, but it is curretnly being worked on in case you did not know.
This is explained step-by-step in The Official Samba-3 HOWTO and Reference Guide. In particuliar, have a look at chapter 21 if you are trying to use your Active Directory as a source for Unix account (ssh, mail, etc). You could also use a combination of nss_ldap, pam-ldap and MS Service For Unix if this is what you are trying to achieve. Note that this is definitely not a project for beginner; it touch some pretty advanced Linux sysadmin-fu like the nsswitch library and PAM.
You are an interesting and insightful person. I volunteer to buy you a keyboard with working Shift keys; it would make reading your intervention so much more enjoyable.
Slashdot Mirror
Maintenance of this cruft certainly have a cost. There might even be economy in using smaller Flash chip for the BIOS.
The WRT54G firmware is not a BIOS, it is more like the complete OS of an embedded system. Below this firmware is another layer of software (I can't recall the name) whose job is to boot the firmware via tftp if boot_wait is set, for example.
Really, I would not want my BIOS to include a shell and a web server !
The problem with the Maximalist approach as you define it is that you will need to update the firmware every time you install a new piece of hardware. I can't believe my firmware will support every sound card and TV tuner under the sun, for example.
I am not an engineer, but I would go with a Minimalist approach augmented with just enough smart to abstract access to the boot device. Which is, in itself, pretty big if you take into account booting from the network, USB/Firewire devices, etc. From there on, the bootloader or the OS kernel should take control.
Some onboard boot code is certainly required, but I would like the BIOS to stay as minimal as possible. Basically, just enough to load the bootloader.
There is trend toward larger and more featureful BIOS. I just don't understand that. Everything that can be done by the OS should be. Updating an OS is much easier than a BIOS, for starter.
And while you are there, why not drop the DOS-supporting cruft ? (like most of the BIOS software interrupts, for example)
Except that you can't use encryption on your link. Bye bye WEP, WPA, ssh, SSL, PGP etc. I am not even sure you could use simple challenge-based authentication scheme such as APOP. Between broadcasting my passwords for anybody to fetch with 20$ worth of equipement or restricting myself to 4 watts EIRP, the choice is an easy one.
73 de VA2***
(sorry for the incomplete callsign, I very much value my privacy on this board)
That is why you need a Senao 2511 or a SMC EliteConnect Wifi adapter. At 200 mW, your signal will actually be *stronger* than the AP.
No. Business is about making money. The "lubrication that prevents the gears of democracy from locking up" is citizen implication and conscientisation. Business does pretty well these day while the current state of Western democracy is damn pathetic. Connect the dots.
A lot of free-market idealists complain about a public administration using taxes to fund a service competing with established business. I personnally won't shed a tear over poor telcos losing a little piece of their near monpo^W^W market, but I can see the point.
...
What would you think, then, of a community network providing free wireless Internet, ran by volunteers and funded with contributions ? Would that still be unfair competition ?
Think NY, Seattle, Austin, Montreal
"Limitless fund and manpower" certainly do not describe most municipal administration.
Ha, you were'nt talking about municipal administration, you where talking about telcos ? Then I see your point !
I do not care about the fate of ISP. Network connectivity is fast becoming a commodity, so most are doomed anyway (only the largest ones will survive in the mid-term). What I do care about is ubiquitous and cheap (if not free) access to the Internet. If municipal WISP can provide that in a cost-effective fashion, why not ?
A WRT54G can be had for 50~80$. It's actually cheaper than most cable modem It can connect all the computers, laptop or desktop, you want.
Considering WiFi networks are just one type of device running on 2.4 Ghz, expect many false positive. In a urban setting, the ISM band is pretty much saturated so your LED will most probably stay a solid green.
Also, I have doubts about the sensitivity of their antenna design.
Not necessarily, but there is plenty to hate with MS beside their files formats.
Keyword here being "most". I am not "most" people, I am a hacker and a thinkerer. The rest of the world can be happy with black box that "just work", and that is fine by me. But the mere knowledge that I hold in my hand a microprocessor-based device that I can't reprogram nag me in some very annoying way. YMMV. Mere mortal can't understand.
We can see that the hacker spirit have definitely left this board when most answers to such a question is "Why would you want to do that anyway ?".
Claiming the bounceback of IT based on the success of a few gadgets and the stock price of Google seem a bit exagerated. What about spending in network infrastructure and large-scale ERP deployement ? That's what IT is as far as I am concerned.
I bet you talk about the change in the c-client library that broke PHP's imap_open() with the /nottls option. You gotta admit that this is a long time ago, and not even on Fedora either.
Regarding Debian, you are right. But I can't help to find it pathetic that the default kernel of the stable distribution, shipped in 2004, still use kernel 2.2. And you can get the same level of version stability by going with a RHEL clone such as CentOS.
I have witnessed a lot of distro snobbism coming from the Debian crowd lately, especially toward Fedora. Debian might be the best thing since sliced-bread, but it also have a lot of stuff hindering its adoption. The Debian development and release process is pretty counter-intuitive to the laymen, and the stable version is horribly outdated. Actually using Debian also require quite a bit of Debian-specific knowledge (ie I never heard about cron-apt and list-bug before), the default install pretty much suck (no less, elvis for vi, etc), the installer is reminding of 1997, etc. Nothing show-stopping by itself here, but a lot of small hinderance. For beginner, the path of least resistance is often one of the RedHat-derived distro, and Fedora is the best known of these.
With Debian, you will end up using either a distribution based on kernel 2.2, or one labelled "testing". You and I know that Debian testing is pretty solid, and get fix diligently when it break. But so does Fedora.
If you where any smarter, you would know how to spell "Operating System".
Wow! I wish FireFox was that innovative !
More power to you if you like Eudora. I don't. We did a very large scale mail server installation (85K accounts) in an organisation that used to have Eudora as the standard MUA. A lot of users of Eudora 5 in the wild. Like everything, bug happen and get corrected so I should not judge the actual quality of a software by it's older releases, but some of them are just too good to pass.
For example, we had to disable STARTTLS IMAP extension because an older release of Eudora was sending STARTSSL. Yes, you read that correctly : STARTSSL. Can you believe that ? That such an obvious bug got passed QA and release astound me. The guy who wrote the STARTTLS support obviously never even bothered to *test* his code.
Another huge source of fuck-up related to Eudora was the habit it had of sending raw 8 bits character in header. In case you don't know, that's a pretty big violation of RFC822. This is particularly thorny as there is no way to tell which charset was used, so automated conversion was not really possible. Did they fixed this one at last ?
There are more I can't remember on the top of my head. Eudora have won the #1 spot in my list of broken MUA thanks to that, above Outlook and OE (who have some interesting way to fuck up too).
Lot's of people in this thread are saying that chead flatbed scanner are really bad for scanning slides. I have a pile of a few hundreds 5x7 family pictures I would like to scan eventually. Budget is very limited, and Linux support is a MUST (no Windows here). I don't necessarily need pro-level or prepress quality, but the nicer the better. I thought about either a Canon CanoLide 20 or 30, as they are cheap and well supported in Linux. How well should they fare ? Any other suggestions ?
ASN.1 is a biggy, but it does'nt stop there. Filter parsing, ACI/ACL, the storage mechanism plus all of the network protocol itself. Hopefully, you will just fork OpenLDAP instead of writing from scratch. Samba is a nice project, but it still require a lot of work. I'd rather see you guys woking on QA and solving exisiting problems.
Whatever float your boat, I guess.
This strike me as being a good example of wheel-reinventing and NIH syndrome. And reinventing a pretty complex wheel at that; LDAP is lightweight only in name, that's for sure. Would'nt it be easier to add the required functionnality and behavior to OpenLDAP, maybe as a plugin/module, a build option or as a runtime configuration directive ?
LDAP is used to centralize information. As a Samba user, I would prefer Samba to use my existing information store if at all possible. Being forced to build a separate DIT just for my Samba stuff is not exactly a rejoicing thought and imply a lot of duplication I am trying to avoid in the first place by using LDAP.
Any pointer toward good discussion about why existing LDAP daemon can't be used for AD interop ? I guess ACI support on individual object is a big one, but it is curretnly being worked on in case you did not know.This is explained step-by-step in The Official Samba-3 HOWTO and Reference Guide. In particuliar, have a look at chapter 21 if you are trying to use your Active Directory as a source for Unix account (ssh, mail, etc). You could also use a combination of nss_ldap, pam-ldap and MS Service For Unix if this is what you are trying to achieve. Note that this is definitely not a project for beginner; it touch some pretty advanced Linux sysadmin-fu like the nsswitch library and PAM.
You are an interesting and insightful person. I volunteer to buy you a keyboard with working Shift keys; it would make reading your intervention so much more enjoyable.