A "real" content management system, and I'm not sure Zope really cuts the mustard (but I don't know), is something Linux in general has been lacking.
It might be worth your while to actually give Zope a roll, as it seems to have *more* features than the go.com stuff. Not sure why you haven't already, as you seem very upbeat about content management systems.
Contrast this with Zope: Browse the UML model for ZODB, and you'll find a DB class, a Connection class, a Transaction class, a Persistent class, but no DiscussionForum class or ForumPosting class. That's because it's one of the very most bottom layers of Zope. You wouldn't put tires on a gas tank, would you? OTOH, browse the UML model for the Zope Portal Toolkit (it's in the file), and you will see things just like you're talking about. I agree, however, that Zope needs more high-level apps built on top of it. We're hoping the community continues to contribute Zope products and we're obviously working on a few as well:-)
Designing a heavily database-driven site which uses URLs without query strings is very possible.
One of the unofficial mottoes of Zope is that it gives you "URLs you can read to your mother." This sets it apart from other app servers like Vignette Story Server, et. al. and many home-grown Perl solutions.
In Zope, there is really no such thing as a static page. Straight HTML pages are stored as Python objects in an object database and are rendered on-the-fly just like any other object.
You don't need to tack all that stuff on to the end of your URL. Really.
As a nonconditional requirement during an interview process, I was required to take this test.
It tries to divide people into 16 "variant temperaments" by evaluating answers to the questions on the test.
Its interesting. I would imagine that most people using Slashdot will end up in the "Rational" group, whereas many people running government would end up in the "Guardian" group. It doesn't presume to rate the intelligence of people, just their temperaments.
It may be that the folks that choose IT as a career fall more into a specific temperament than those who go into, say, automotive engineering.
I developed the basic framework of our corporate intranet in about 40 hours using Zope. Know that I went from "ground zero" knowledge of Zope and/or Python to fully functional site in this time. Our intranet current performs the following:
- Provides dummyproof interface to upload and download files per-department. They can also delete their uploaded files. Security is managed on a per-department basis. Users can manage their own department's user permissions. They can also maintain a private filestore.
- Allows users to add news items in their respective which are displayed on the home page. Users may also edit and delete their own news items.
- Allows our "partnership coordinator" to post information about our various industry partnerships (we're an IT consultancy, as such we have relationships with many vendors, etc.)
- Allows users to search all text-based files on the site for specific words contained in the id, title, or content.
It's not done, but it's getting there. I posted an analysis of my experiences with both Zope and Microsoft Site Server in this role: http://sharon.iqgroup.com/zopevssiteserver.html
MiniVend is an excellent piece of software. This is a big market for Linux and related open-source packages. It makes perfect sense for RedHat to bundle and package this. It makes my job easier, too, because when people ask me what it runs on, I can say "RedHat E-Commerce Server" instead of getting down in the mire of having to explain and defend the open-source mentality. People want to pay for this stuff. This is a good thing.
I ran a BBS on a C64 (and later an Amiga and even later an XT) for years named the Enchanted Toilet in Southern New Jersey, US. The original BBS on C64 had two whopping single-sided 1541 5.25" floppy disk drives, and the BBS (C-Net) was written in BASIC. It rocked. The Amiga version ran on Citadel-68K and had a wire-wrapped HDD interface built from a kit that drove a humongous 30MB RLL Seagate ST238 disk drive. I later cheated Seagate by adding an ST225 (MFM) in RLL mode to get 30MB out of a 20MB HDD. This setup also rocked. The XT version also ran on Citadel (Cit/86), but who cares about hardware on a PC-compatible.. I don't even remember what it had.
Christ, who knew about the Internet? We were just happy to get Atari ST's and Apple II's talking to one another then... I guess I lifted a couple passwords and dialed in to Rutgers to get on IP-connected MUDs, but what the hell did I know? All I knew is that this stuff was kick-ass. And as you can tell, they're still trying to pry me out of my cathode-ray-basked desk chair today.
Actually, if you're talking about a Linux "agent" for Veritas (formerly Seagate) Backup Exec, it exists currently. I've just finished installing it on 5 Linux machines we have here. It's (obviously) not open source.
To do this I hadx to buy the 7.3 version of Backup Exec for NT... (I upgraded from 7.0... it was like $400 for the upgrade). On the CD is a Linux "agent". It installs pretty nicely if you have Red Hat 5.2... otherwise gives you a warning about "unsupported UNIX platform" but worked OK on RH 5.0 and RH 6.0 (we don't have any other distributions, call me a follower... my guess is that it wont be pleasant to get it to work under another distro) It so far is also a real pig... Linux clients get backed up to the NT BE server at about 4.8MB/min (what is that? 500kbps?). In any case, 4.8MB/min across an Ethernet during a backup is not good. By way of comparison, our Windows clients get backed up at about 170MB/min.
Additionall problems: the agent requires a password for access by the backup server... you specify the password during the install of the agent on the Linux machines.
The password is:
1. shown in the clear as you type it in (the install is just a batch script, guess they didn't know how to change the tty settings during the password entry). 2. stored in the clear in a config file 3. exposed to all users, as the directory the agent install creates and the agent config file is permitted rx for other. (i changed it, of course, to be unreadable by anyone but root)
three things that tip you off that the guys in this division of Veritas dont write much UNIX software:
1. the speed 2. the default security problems i mentioned 3. the agents come in a tar file that, when untarred,expands files into the current directory instead of creating a directory from your current one (not a big deal, but its a tipoff that they're used to PKZIP). 4. the post-agent-install tip that you can start the daemon by "restarting your workstation". so funny. not just "/etc/rc.d/init.d/agent.init start", but "restart your workstation". it's classic.
In any case, I'm glad the agent exists even with its faults because it makes my life much easier.. rather than having to administer 2 backup systems (one for NT, one for *NIX), I can do it all from one console... Of course, I don't know if it actually completely *works* yet, as the backup is running as we speak, but hopefully the verify results work out...
Bottom line observation: Commercial software running on Linux is sort of scary. None of the niceties of community-produced stuff. I dont think I'd *want* to see the code...
This is taken from a diff that makes Whois.pm work again. It is against Whois.pm 1.13. Insert these lines after line 200.
(There's a variable ($text) that gets the info from a socket. Just substitute null for the whole disclaimer phrase with a regex assign.)
201,205c201 my $disclaimer; $disclaimer = EOF; Access to Network Solutions' WHOIS information is provided to assist persons in determining the contents of a domain name registration record in NSI's registrar database. The data in this record is provided by NSI for informational purposes only, and NSI does not guarantee its accuracy. Compilation, repackaging, dissemination, or other use of the WHOIS database in its entirety, or a substantial portion thereof, is not allowed without NSI's prior written permission. By submitting this query, you agree to abide by this policy. All rights reserved. EOF $text=~s/$disclaimer//;
If this is an issue, keep your user authentication database wherever it is currently (Novell, NT, RADIUS, TACACS+, what have you) and use PAM (pluggable authentication modules) under Linux to authenticate users against the database there.
It works very well. We use PAM_SMB here and as a result all the users have the same UNIX password as NT password, though we must administer the usernames semi-manually on the UNIX side.
See http://www.pmg.com for a protocol decode file conversion utility that will allow you to bring tcpdump output (among others) into it and write it as Sniffer or NetXRay or what-have-you formats.
OK. I admit it. I'm a Neal Stephenson zealot. I've read Snow Crash. I've read Snow Crash *aloud*. To other people. Without their consent. Zodiac, Diamond age, ingested.
But seeing that essay linked through Slashdot is really a kicker. Finding out Neal Stephenson is as oddly fixated on Linux as I am is like walking into my regular corner bar and finding out that, previously unbeknownst to anyone, Metallica is playing there that night because they know the owner from years ago and by the way James Hetfield might need to crash at my place tonight if its OK with me.
I love it.
And you know you're reading this, too, Neal. No self-respecting Linux pseudogeek does not read Slashdot at least on occasion. And no writer is enough lacking in vanity to not read his critics.
You, sir, are the tits. Good luck with the new book.
Slashdot Mirror
A "real" content management system, and I'm not sure Zope really cuts the mustard (but I don't know), is something Linux in general has been lacking.
It might be worth your while to actually give Zope a roll, as it seems to have *more* features than the go.com stuff. Not sure why you haven't already, as you seem very upbeat about content management systems.
> I'm still waiting for A Scanner Darkly... movie.
I second that. That book blows DODOES away by a long shot.
Contrast this with Zope: Browse the UML model for ZODB, and you'll find a DB class, a Connection class, a Transaction class, a Persistent class, but no DiscussionForum class or ForumPosting class. That's because it's one of the very most bottom layers of Zope. You wouldn't put tires on a gas tank, would you? OTOH, browse the UML model for the Zope Portal Toolkit (it's in the file), and you will see things just like you're talking about. I agree, however, that Zope needs more high-level apps built on top of it. We're hoping the community continues to contribute Zope products and we're obviously working on a few as well :-)
Designing a heavily database-driven site which uses URLs without query strings is very possible.
One of the unofficial mottoes of Zope is that it gives you "URLs you can read to your mother." This sets it apart from other app servers like Vignette Story Server, et. al. and many home-grown Perl solutions.
In Zope, there is really no such thing as a static page. Straight HTML pages are stored as Python objects in an object database and are rendered on-the-fly just like any other object.
You don't need to tack all that stuff on to the end of your URL. Really.
It tries to divide people into 16 "variant temperaments" by evaluating answers to the questions on the test.
Its interesting. I would imagine that most people using Slashdot will end up in the "Rational" group, whereas many people running government would end up in the "Guardian" group. It doesn't presume to rate the intelligence of people, just their temperaments.
It may be that the folks that choose IT as a career fall more into a specific temperament than those who go into, say, automotive engineering.
I developed the basic framework of our corporate intranet in about 40 hours using Zope. Know that I went from "ground zero" knowledge of Zope and/or Python to fully functional site in this time. Our intranet current performs the following:
- Provides dummyproof interface to upload and download files per-department. They can also delete their uploaded files. Security is managed on a per-department basis. Users can manage their own department's user permissions. They can also maintain a private filestore.
- Allows users to add news items in their respective which are displayed on the home page. Users may also edit and delete their own news items.
- Allows our "partnership coordinator" to post information about our various industry partnerships (we're an IT consultancy, as such we have relationships with many vendors, etc.)
- Allows users to search all text-based files on the site for specific words contained in the id, title, or content.
It's not done, but it's getting there. I posted an analysis of my experiences with both Zope and Microsoft Site Server in this role: http://sharon.iqgroup.com/zopevssiteserver.html
Zope really shines. It's really just great.
I am using Zope. Rocks. Our intranet now runs on Zope, and soon a customer's.
You're too late. It already is.
MiniVend is an excellent piece of software. This is a big market for Linux and related open-source packages. It makes perfect sense for RedHat to bundle and package this. It makes my job easier, too, because when people ask me what it runs on, I can say "RedHat E-Commerce Server" instead of getting down in the mire of having to explain and defend the open-source mentality. People want to pay for this stuff. This is a good thing.
I ran a BBS on a C64 (and later an Amiga and even later an XT) for years named the Enchanted Toilet in Southern New Jersey, US. The original BBS on C64 had two whopping single-sided 1541 5.25" floppy disk drives, and the BBS (C-Net) was written in BASIC. It rocked. The Amiga version ran on Citadel-68K and had a wire-wrapped HDD interface built from a kit that drove a humongous 30MB RLL Seagate ST238 disk drive. I later cheated Seagate by adding an ST225 (MFM) in RLL mode to get 30MB out of a 20MB HDD. This setup also rocked. The XT version also ran on Citadel (Cit/86), but who cares about hardware on a PC-compatible.. I don't even remember what it had.
Christ, who knew about the Internet? We were just happy to get Atari ST's and Apple II's talking to one another then... I guess I lifted a couple passwords and dialed in to Rutgers to get on IP-connected MUDs, but what the hell did I know? All I knew is that this stuff was kick-ass. And as you can tell, they're still trying to pry me out of my cathode-ray-basked desk chair today.
Oops... actually, the config file *is* protected with mode user read-only.... my bad. Sorry Veritas guys.
Actually, if you're talking about a Linux "agent" for Veritas (formerly Seagate) Backup Exec, it exists currently. I've just finished installing it on 5 Linux machines we have here. It's (obviously) not open source.
To do this I hadx to buy the 7.3 version of Backup Exec for NT... (I upgraded from 7.0... it was like $400 for the upgrade). On the CD is a Linux "agent". It installs pretty nicely if you have Red Hat 5.2... otherwise gives you a warning about "unsupported UNIX platform" but worked OK on RH 5.0 and RH 6.0 (we don't have any other distributions, call me a follower... my guess is that it wont be pleasant to get it to work under another distro) It so far is also a real pig... Linux clients get backed up to the NT BE server at about 4.8MB/min (what is that? 500kbps?). In any case, 4.8MB/min across an Ethernet during a backup is not good. By way of comparison, our Windows clients get backed up at about 170MB/min.
Additionall problems: the agent requires a password for access by the backup server... you specify the password during the install of the agent on the Linux machines.
The password is:
1. shown in the clear as you type it in (the install is just a batch script, guess they didn't know how to change the tty settings during the password entry).
2. stored in the clear in a config file
3. exposed to all users, as the directory the agent install creates and the agent config file is permitted rx for other. (i changed it, of course, to be unreadable by anyone but root)
three things that tip you off that the guys in this division of Veritas dont write much UNIX software:
1. the speed
2. the default security problems i mentioned
3. the agents come in a tar file that, when untarred,expands files into the current directory instead of creating a directory from your current one (not a big deal, but its a tipoff that they're used to PKZIP).
4. the post-agent-install tip that you can start the daemon by "restarting your workstation". so funny. not just "/etc/rc.d/init.d/agent.init start", but "restart your workstation". it's classic.
In any case, I'm glad the agent exists even with its faults because it makes my life much easier.. rather than having to administer 2 backup systems (one for NT, one for *NIX), I can do it all from one console... Of course, I don't know if it actually completely *works* yet, as the backup is running as we speak, but hopefully the verify results work out...
Bottom line observation: Commercial software running on Linux is sort of scary. None of the niceties of community-produced stuff. I dont think I'd *want* to see the code...
I can't make the code format correctly in a reply.
2-minute quick fix.
This is taken from a diff that makes Whois.pm work again. It is against Whois.pm 1.13. Insert these lines after line 200.
(There's a variable ($text) that gets the info from a socket. Just substitute null for the whole disclaimer phrase with a regex assign.)
201,205c201
my $disclaimer;
$disclaimer = EOF;
Access to Network Solutions' WHOIS information is provided to assist persons in determining the contents of a domain name registration record in NSI's registrar database. The data in this record is provided by NSI for informational purposes only, and NSI does not guarantee its accuracy. Compilation, repackaging, dissemination, or other use of the WHOIS database in its entirety, or a substantial portion thereof, is not allowed without NSI's prior written permission. By submitting this query, you agree to abide by this policy. All rights reserved.
EOF
$text=~s/$disclaimer//;
Nailed me too. I'm going through the source of Whois.pm trying to figure out where to start. I will let you know if I get somewhere.
Why do I always seem to be fixing stuff that other people break?
Oh wait... that's my job.
If this is an issue, keep your user authentication database wherever it is currently (Novell, NT, RADIUS, TACACS+, what have you) and use PAM (pluggable authentication modules) under Linux to authenticate users against the database there.
It works very well. We use PAM_SMB here and as a result all the users have the same UNIX password as NT password, though we must administer the usernames semi-manually on the UNIX side.
See http://www.pmg.com for a protocol decode file conversion utility that will allow you to bring tcpdump output (among others) into it and write it as Sniffer or NetXRay or what-have-you formats.
OK. I admit it. I'm a Neal Stephenson zealot. I've read Snow Crash. I've read Snow Crash *aloud*. To other people. Without their consent. Zodiac, Diamond age, ingested.
But seeing that essay linked through Slashdot is really a kicker. Finding out Neal Stephenson is as oddly fixated on Linux as I am is like walking into my regular corner bar and finding out that, previously unbeknownst to anyone, Metallica is playing there that night because they know the owner from years ago and by the way James Hetfield might need to crash at my place tonight if its OK with me.
I love it.
And you know you're reading this, too, Neal. No self-respecting Linux pseudogeek does not read Slashdot at least on occasion. And no writer is enough lacking in vanity to not read his critics.
You, sir, are the tits. Good luck with the new book.
Many thanks,
Chris