I have to agree here. I have been running a RPM based distro (Caldera) and most of the binary RPMS out there are very RH specific. Things are packaged differently, installed in different places, called different names or the worst--dynamically linked to a specific version of a library. I only download source SRPMS or tarballs (thanks setup.sh), its the only way that I can be sure of correct dependancys. Even then it doesn't work if the package requires distro specific tools to install, that I don't have (chkconfig comes to mind).
I don't find it fun to spend days chasing down library dependancies and trying to integrate them without breaking my system. I tried to install GNOME when it came out but instead spent days in DLL hell trying to get and compile every special version of the libraries for librarys that every other library needed. Some people get off on spending days futzing with stuff just to get it to compile, I now compile all the software I download, but I actually want to USE the stuff in my lifetime.
(Wake up on the wrong side of nobody this morning?)
Re:I had problems with the RPM for Rh 6.0/6.1
on
Samba 2.06 Released
·
· Score: 1
Have the best of both worlds, use setup.sh it builds and RPM out of almost any GNU Autoconf based tarball. That way you can keep track of what you have installed, and where it is.
That's why I have never understood why GTK tries to emulate the Motif look/feel. Why didn't they use something more NeXTalike, GNU already has the GNUStep project and WindowMaker is the official WM for it. Everybody liked the NeXT interface, I don't see what benefit they have from going their own way--except to be anti-KDE/QT.
QT would make sense since it is available for Windows as well as Unix. They could relink all their programs to QT and have instant portability with one codebase while still keeping the Windows versions. Since they are using KDE extensively for their Corel Linux, and Office will probably be a big draw for that platform, it would be a no-brainer.
Corel Office for Java (I might still have a beta on Zip disk somewhere) would have been a good product but it was a bit ahead of the times (Moore's Law). Hopefully they still have the sources kicking around somewheres, to compete with the Web versions of Sun StarOffice and MS Office. While it does seem like a dumb idea, with the current state of bandwidth, to do this it might be a buzzword and checkbox item in the future. They might get left out, labled as an also-ran, even if the technology is pointless (think Push/Channels)
In the future, when any two arbitrary points on the Internet have the bandwidth of a T3, then it will be feasable to centralize and have whole Office suites in Java. But it is pretty pointless now.
Disregard the previous message, I just hopped on over to the Blackbox website and saw that they have a newer version. The changelog states that they have applied a fix to this particular problem, I am recompiling now. BTW, thank god for setup.sh, it is automatically building a RPM from the tarball I downloaded to replace the Blackbox RPM that I have installed currently. Y'all should try it, setup.sh builds RPMS from any tarball that is built using GNU Autoconf and has a LSM file.
I would like to try it to but their script crashed when I hit submit in KFM. I tried to go back and resend the form but got a MS ODBC error. I tried to go to the top level of the server and work my way down but they haven't even gotten rid of the Netscape Enterprise Server default homepage. Then I tried to complain about it on Slashdot last night and Blackbox WM crashed. Arrrgggghhhhh!!!!!
PS and By-The-Way: Anyone else having a problem with Blackbox & Netscape. Whenever I get a logon popup box in Netscape, Blackbox crashes taking every other window with it and dumping me back to KDM.
Try looking at CUPS the Common UNIX Printing System. It aims to remove many of these shortcommings. CUPS is supposed to be able to give printer capability information back to the application, like every other OS, so that it is easy to change from Portrait to Landscape or whatever. It is compatable with BSD and SysV printing systems but its native protocol is the new IPP Internet Printing Protocol. I haven't had time to play with it but it looks like it should take Unix printing at least into the 90's.
Re:I had problems with the RPM for Rh 6.0/6.1
on
Samba 2.06 Released
·
· Score: 1
Actually it can be pretty simple. Log in as root and run "rpm --rebuild package-1.2.3-4.src.rpm". It will untar the files (in/usr/src/$DISTRONAME/SOURCES) and compile (in/usr/src/$DISTRONAME/BUILD) them based on a script (in/usr/src/$DISTRONAME/SPECS). If the compile doesn't work right (like it doesn't use the Autoconf variables you want) just edit the spec file in./SPECS and run "rpm -bb./SPECS/packagename.spec" to build binary (bb) packages. Simple.
>And no, the FBI does not have the manpower to listen into your phone unless you are quite the bad*ss. Even then, a federal judge has to approve the warrant (the legal document, not the band) that will allow them to wiretap you. The reason that we get so uptight about these things is that they DO have the power to monitor people wholesale (Echelon, etc.) Anything that makes it easy for them to wiretap makes it easy for them to abuse their privelige. Do you really believe that most wiretaps go through a Judge first? In LA recently there is a big stink because the police were holding massive wiretaps using a few old warrents, or no warrent at all. Even if the data they collected wasn't admissible in court it allowed them to intimidate suspects and investigate individuals without proper probable cause. If it is easy and automatic for them to wiretap do you think they will hesitate to use their power on a whim? I don't trust the Government that far. Remember, when you give away power you are giving it to the lowest common denominator. It doesn't matter how many right and just officers would use this technology, one fuck-up loser can do alot of damage. I don't want to give that loser that kind of power. I don't think that this IETF proposal will come to anything, it is against the principles that they operate under. Even if they did recommend a wiretapping protocol it probaly wouldn't go anywhere, we would just go through, under and around it. This only works, however, if we actually do it. Reading the post about the deCSS debacle the other day I noticed a number of "I didn't download the source and now it is gone, help!" posts. What if no one, or very few, had hosted the source elsewhere, deCSS would be gone and we would have to recode it (a simplistic example, many people d/l the source to compile it on their systems). I also remember the GLDoom port that died after over a year of work because of HD failure and lack of backups. Don't let this happen to our Internet!
You can replace/recharge the batteries you know. Having all that music on the player means less time transferring to and from your desktop machine. Who wants to spend all their days managing the memory and playlists for a machine that only holds a few songs? And of course there is the "Jukebox" aspect of it. With the ability to host music wholesale you can pick and choose the mix without worrying if you downloaded that particular song or not. With that much space it could conceviably be your main MP3 storage device.
You are making the assumption that if RHAT buys Cygnus than they will be dropping support for other OSs. I really doubt that could happen. Remember Cygnus is the maintainer of GCC/EGCS, all their software is GPLd. The same with RH, all the stuff they create in-house is GPLd too.
This is probably a good idea, get Cygnus in contact with some of that IPO money before it runs out. And aquire a small stable buisness at the same time.
I doubt that Cygnus would stand for much bullying on RedHats part, and they could always take the code with them. Remember, traditional software paradigms don't work in a GPL environment.
Try removing every device from Device Manager before you upload the disk image. Then when you install it every hardware device will be detected upon bootup. Not a perfect solution, relies on Windows hardware detection, but it could be workable. Course of you aren't using PnP NICs and things this could be more difficult. I suppose that a master batch file, that only runs upon first boot and removes itself, could run anything needed to be customized on a per workstation basis.
That is such a sad story. I have had better luck. Well they still forgot to enable DMA and a few other niggling things but nothing so bad as your shop. At the tech school where I administer we have a Ghost image for each computer lab. When Windows decided to crap out I just walk in with the boot disk and walk out 5 min later, come back to change the workstation name so that it shows up correctly in ZEN and that's it. It sure beats trying to figure out why a particular install went bad, takes about 30 min on the buildings network.
But yes, if an idiot creates the initial install images, then every computer in your org will be screwed up in the same way (this can actually help troubleshooting, every system breaks the same, and can be fixed the same). Conversely if the initial image is good than every computer will be that much more stable. It also helps when you want to distribute software through a logon script, you can be sure that every computer has the same software installed in the same places.
These tools are much better than trying to create custom automated install scripts and response files. It is faster and requires less babysitting.
See I think that the Linux autoprobing code is a bit of a kludge. Every module has its own autoprobing and they sometimes step on each other causing the computer to lock. Usually this is only a problem during install when everything is being probed for. Also several modules do not detect their hardware when it is installed in the machine and have to be told manually where it is. While I should be able to force a configuration if I want, autoprobing should be able to find the signatures, PnP ids and PCI ids of everything and match it up against a central database. Devices like my Gravis UltraSound, aha152x SCSI controller and NE2000 compat ethernet have never been autodetected by Linux, I always have to load the modules manually with specific settings after install. The Add New Hardware Wizard can autodetect these by itself pretty easily.
I do agree that MS design and MS implementation are two wildly different beasts. They couldn't code their way out of a paper sack, even with a pointer out.
At a place where I used to work we had the PC optical mouse, as well as a couple of SPARCStations. It was pretty neat to have the same technology on advanced workstations as well as on PCs. Sorry to hear that the Mouse is dead, I personally own a ProAgio and am very happy with it.
While their scroll wheel mouse is pretty good, and their marketing was able to make it very popular they weren't the first. I have an old Mouse Systems mouse that has 5 buttons and a scroll wheel that is older than the IntelliMouse. It is actually better for me because the scroll wheel is free-moving, it doesn't have set stops like the MS and Logitech mice. It is a real tank too, I've used it for several years and it still tracks beautifully.
Only problem is that the wheel doesn't seem to work in X, no one seems to have heard of the ProAgio and the proprietary MouseSystems protocol that it talks. In fact it doesn't work in Windows either (the entire mouse, not just the wheel) without drivers from MouseSystems, even though the MS protocol has been around almost as long as mice.
> There's no innovation in windows, all the Linux community should care about is the hardware support.
While that is a popular opinion on Slashdot it is not entirely true. There are several features that I would like to rip from Windows if given the chance. The people who designed Windows (and NT) are not complete idiots as some would have us believe. MS has one of the largest concentrations of CompSci PhDs anywhere. Of course this hasn't seemed to help their code quality, every time they start a project they try to make it the most whizz bangy thing and then fail.
Things I would rip from Windows:
1) Add New Hardware Wizard 2) INF hardware datafile reader 3) DirectX (yeah I know D3D sucks, the rest is pretty decent though) 4) Explorer file manager (although KFM v2 will be better) 4) Device Manager control dialogs 5) Various other control panels, like the Network panel and UI bits (MS has spent alot of time on UI research and have many design wins, anyone notice how ktop looks exactally like WinNT task manager, only done right)
There is a few other things that I disremember now but the point is that just because it came from MS doesn't automatically make it shoddy and worthless. I mean, look at their hardware division, they make great stuff. The KDE project has done well in ripping off good ideas from several OSs, like Mac, Windows, CDE (I know this isn't an OS so sue me), etc.
This shouldn't be two hard to implement, just make a clone of or license someone elses boot prom. Like Apples FORTH interpreter or something. Start putting this on new PCI only boards, the ones without any serial/parallel/ps2 ports. There backwards compatability isn't a problem, you only need limited support from some popular OSs (Windows9x is really the only one that uses the BIOS for much of anything). Maybe you can even eschew Win9x compatability seeing that Win2K, BeOS, Linux, etc would be available at the time.
I still have the swh.exe Hercules mono version of SpaceWar around somewhere. Great fun, although I had a black and white monitor. I am really sorry that I threw it out several years ago, it would make a great terminal for my Linux box, or I could have dual monitor DOS on my old 386-16 1.5MB RAM that is two small to run Linux.
Slashdot Mirror
I have to agree here. I have been running a RPM based distro (Caldera) and most of the binary RPMS out there are very RH specific. Things are packaged differently, installed in different places, called different names or the worst--dynamically linked to a specific version of a library. I only download source SRPMS or tarballs (thanks setup.sh), its the only way that I can be sure of correct dependancys. Even then it doesn't work if the package requires distro specific tools to install, that I don't have (chkconfig comes to mind).
I don't find it fun to spend days chasing down library dependancies and trying to integrate them without breaking my system. I tried to install GNOME when it came out but instead spent days in DLL hell trying to get and compile every special version of the libraries for librarys that every other library needed. Some people get off on spending days futzing with stuff just to get it to compile, I now compile all the software I download, but I actually want to USE the stuff in my lifetime.
Enough ranting from the Peanut Gallery.
Yeah, whatever you say, Metcalf.
Well, at least nothing bad happens!
(Wake up on the wrong side of nobody this morning?)
Have the best of both worlds, use setup.sh it builds and RPM out of almost any GNU Autoconf based tarball. That way you can keep track of what you have installed, and where it is.
That's why I have never understood why GTK tries to emulate the Motif look/feel. Why didn't they use something more NeXTalike, GNU already has the GNUStep project and WindowMaker is the official WM for it. Everybody liked the NeXT interface, I don't see what benefit they have from going their own way--except to be anti-KDE/QT.
QT would make sense since it is available for Windows as well as Unix. They could relink all their programs to QT and have instant portability with one codebase while still keeping the Windows versions. Since they are using KDE extensively for their Corel Linux, and Office will probably be a big draw for that platform, it would be a no-brainer.
Corel Office for Java (I might still have a beta on Zip disk somewhere) would have been a good product but it was a bit ahead of the times (Moore's Law). Hopefully they still have the sources kicking around somewheres, to compete with the Web versions of Sun StarOffice and MS Office. While it does seem like a dumb idea, with the current state of bandwidth, to do this it might be a buzzword and checkbox item in the future. They might get left out, labled as an also-ran, even if the technology is pointless (think Push/Channels)
In the future, when any two arbitrary points on the Internet have the bandwidth of a T3, then it will be feasable to centralize and have whole Office suites in Java. But it is pretty pointless now.
Enough babbling from myself.
Disregard the previous message, I just hopped on over to the Blackbox website and saw that they have a newer version. The changelog states that they have applied a fix to this particular problem, I am recompiling now. BTW, thank god for setup.sh, it is automatically building a RPM from the tarball I downloaded to replace the Blackbox RPM that I have installed currently. Y'all should try it, setup.sh builds RPMS from any tarball that is built using GNU Autoconf and has a LSM file.
I would like to try it to but their script crashed when I hit submit in KFM. I tried to go back and resend the form but got a MS ODBC error. I tried to go to the top level of the server and work my way down but they haven't even gotten rid of the Netscape Enterprise Server default homepage. Then I tried to complain about it on Slashdot last night and Blackbox WM crashed. Arrrgggghhhhh!!!!!
PS and By-The-Way: Anyone else having a problem with Blackbox & Netscape. Whenever I get a logon popup box in Netscape, Blackbox crashes taking every other window with it and dumping me back to KDM.
Try looking at CUPS the Common UNIX Printing System. It aims to remove many of these shortcommings. CUPS is supposed to be able to give printer capability information back to the application, like every other OS, so that it is easy to change from Portrait to Landscape or whatever. It is compatable with BSD and SysV printing systems but its native protocol is the new IPP Internet Printing Protocol. I haven't had time to play with it but it looks like it should take Unix printing at least into the 90's.
Actually it can be pretty simple. Log in as root and run "rpm --rebuild package-1.2.3-4.src.rpm". It will untar the files (in /usr/src/$DISTRONAME/SOURCES) and compile (in /usr/src/$DISTRONAME/BUILD) them based on a script (in /usr/src/$DISTRONAME/SPECS). If the compile doesn't work right (like it doesn't use the Autoconf variables you want) just edit the spec file in ./SPECS and run "rpm -bb ./SPECS/packagename.spec" to build binary (bb) packages. Simple.
>And no, the FBI does not have the manpower to listen into your phone unless you are quite the bad*ss. Even then, a federal judge has to approve the warrant (the legal document, not the band) that will allow them to wiretap you. The reason that we get so uptight about these things is that they DO have the power to monitor people wholesale (Echelon, etc.) Anything that makes it easy for them to wiretap makes it easy for them to abuse their privelige. Do you really believe that most wiretaps go through a Judge first? In LA recently there is a big stink because the police were holding massive wiretaps using a few old warrents, or no warrent at all. Even if the data they collected wasn't admissible in court it allowed them to intimidate suspects and investigate individuals without proper probable cause. If it is easy and automatic for them to wiretap do you think they will hesitate to use their power on a whim? I don't trust the Government that far. Remember, when you give away power you are giving it to the lowest common denominator. It doesn't matter how many right and just officers would use this technology, one fuck-up loser can do alot of damage. I don't want to give that loser that kind of power. I don't think that this IETF proposal will come to anything, it is against the principles that they operate under. Even if they did recommend a wiretapping protocol it probaly wouldn't go anywhere, we would just go through, under and around it. This only works, however, if we actually do it. Reading the post about the deCSS debacle the other day I noticed a number of "I didn't download the source and now it is gone, help!" posts. What if no one, or very few, had hosted the source elsewhere, deCSS would be gone and we would have to recode it (a simplistic example, many people d/l the source to compile it on their systems). I also remember the GLDoom port that died after over a year of work because of HD failure and lack of backups. Don't let this happen to our Internet!
You can replace/recharge the batteries you know.
Having all that music on the player means less time transferring to and from your desktop machine. Who wants to spend all their days managing the memory and playlists for a machine that only holds a few songs?
And of course there is the "Jukebox" aspect of it. With the ability to host music wholesale you can pick and choose the mix without worrying if you downloaded that particular song or not. With that much space it could conceviably be your main MP3 storage device.
>). If RedHat bought Cygnus, what would WRS do?
You are making the assumption that if RHAT buys Cygnus than they will be dropping support for other OSs. I really doubt that could happen. Remember Cygnus is the maintainer of GCC/EGCS, all their software is GPLd. The same with RH, all the stuff they create in-house is GPLd too.
This is probably a good idea, get Cygnus in contact with some of that IPO money before it runs out. And aquire a small stable buisness at the same time.
I doubt that Cygnus would stand for much bullying on RedHats part, and they could always take the code with them. Remember, traditional software paradigms don't work in a GPL environment.
Try removing every device from Device Manager before you upload the disk image. Then when you install it every hardware device will be detected upon bootup. Not a perfect solution, relies on Windows hardware detection, but it could be workable. Course of you aren't using PnP NICs and things this could be more difficult. I suppose that a master batch file, that only runs upon first boot and removes itself, could run anything needed to be customized on a per workstation basis.
That is such a sad story. I have had better luck. Well they still forgot to enable DMA and a few other niggling things but nothing so bad as your shop. At the tech school where I administer we have a Ghost image for each computer lab. When Windows decided to crap out I just walk in with the boot disk and walk out 5 min later, come back to change the workstation name so that it shows up correctly in ZEN and that's it. It sure beats trying to figure out why a particular install went bad, takes about 30 min on the buildings network.
But yes, if an idiot creates the initial install images, then every computer in your org will be screwed up in the same way (this can actually help troubleshooting, every system breaks the same, and can be fixed the same). Conversely if the initial image is good than every computer will be that much more stable. It also helps when you want to distribute software through a logon script, you can be sure that every computer has the same software installed in the same places.
These tools are much better than trying to create custom automated install scripts and response files. It is faster and requires less babysitting.
Obviously, AC, you're the one everyone is raving about.
Thank you for your support.
Speak for yourself. While I always have the searchlights out for a meaningful relationship with my soulmate, I'll settle for sex.
See I think that the Linux autoprobing code is a bit of a kludge. Every module has its own autoprobing and they sometimes step on each other causing the computer to lock. Usually this is only a problem during install when everything is being probed for. Also several modules do not detect their hardware when it is installed in the machine and have to be told manually where it is. While I should be able to force a configuration if I want, autoprobing should be able to find the signatures, PnP ids and PCI ids of everything and match it up against a central database. Devices like my Gravis UltraSound, aha152x SCSI controller and NE2000 compat ethernet have never been autodetected by Linux, I always have to load the modules manually with specific settings after install. The Add New Hardware Wizard can autodetect these by itself pretty easily.
I do agree that MS design and MS implementation are two wildly different beasts. They couldn't code their way out of a paper sack, even with a pointer out.
At a place where I used to work we had the PC optical mouse, as well as a couple of SPARCStations. It was pretty neat to have the same technology on advanced workstations as well as on PCs. Sorry to hear that the Mouse is dead, I personally own a ProAgio and am very happy with it.
While their scroll wheel mouse is pretty good, and their marketing was able to make it very popular they weren't the first. I have an old Mouse Systems mouse that has 5 buttons and a scroll wheel that is older than the IntelliMouse. It is actually better for me because the scroll wheel is free-moving, it doesn't have set stops like the MS and Logitech mice. It is a real tank too, I've used it for several years and it still tracks beautifully.
Only problem is that the wheel doesn't seem to work in X, no one seems to have heard of the ProAgio and the proprietary MouseSystems protocol that it talks. In fact it doesn't work in Windows either (the entire mouse, not just the wheel) without drivers from MouseSystems, even though the MS protocol has been around almost as long as mice.
> There's no innovation in windows, all the Linux community should care about is the hardware support.
While that is a popular opinion on Slashdot it is not entirely true. There are several features that I would like to rip from Windows if given the chance. The people who designed Windows (and NT) are not complete idiots as some would have us believe. MS has one of the largest concentrations of CompSci PhDs anywhere. Of course this hasn't seemed to help their code quality, every time they start a project they try to make it the most whizz bangy thing and then fail.
Things I would rip from Windows:
1) Add New Hardware Wizard
2) INF hardware datafile reader
3) DirectX (yeah I know D3D sucks, the rest is pretty decent though)
4) Explorer file manager (although KFM v2 will be better)
4) Device Manager control dialogs
5) Various other control panels, like the Network panel and UI bits (MS has spent alot of time on UI research and have many design wins, anyone notice how ktop looks exactally like WinNT task manager, only done right)
There is a few other things that I disremember now but the point is that just because it came from MS doesn't automatically make it shoddy and worthless. I mean, look at their hardware division, they make great stuff. The KDE project has done well in ripping off good ideas from several OSs, like Mac, Windows, CDE (I know this isn't an OS so sue me), etc.
>> i'm persona non grata here. i don't write code. i just build networks.
Not true, oh bandwidth god. Spread some of that joy around.
Hear, hear!
This shouldn't be two hard to implement, just make a clone of or license someone elses boot prom. Like Apples FORTH interpreter or something. Start putting this on new PCI only boards, the ones without any serial/parallel/ps2 ports. There backwards compatability isn't a problem, you only need limited support from some popular OSs (Windows9x is really the only one that uses the BIOS for much of anything). Maybe you can even eschew Win9x compatability seeing that Win2K, BeOS, Linux, etc would be available at the time.
Just my $0.02 US.
I still have the swh.exe Hercules mono version of SpaceWar around somewhere. Great fun, although I had a black and white monitor. I am really sorry that I threw it out several years ago, it would make a great terminal for my Linux box, or I could have dual monitor DOS on my old 386-16 1.5MB RAM that is two small to run Linux.