blue sky thinking

Hi,
 
This is just blue sky thinking and not pertaining to any
institution in
particular.  What if an institution wanted to gain the
benefits that accrue
to libraries that have excellent server access and support
in, say, a LAMP
(Linux / MySQL / Apache / Php & PERL) environment but
had limited means and
possibilities.  Is there any reason why something like this
scenario
wouldn't work?
 
1)  buy cheap space on a Linux shared hosting server that
provides root
access such as 
      http://www.spry.com/pl
esk-vps/ <http://www.spry.co
m/plesk-vps/>  /
htt
p://support.jodohost.com/showthread.php?t=1726
<http://support.jodohost.com/showthread.php?t=1726> 
/
http://www
.linode.com/products/linodes.cfm
<http:/
/www.linode.com/products/linodes.cfm> 
 
2) identify open source tools that significantly expand
capabilities (i.e.
Content Management Systems or other database server driven
tools)
 
3)  either install and configure these tools oneself or
simply outsource it
to someone via a site like http://www.elance.com/
<http://www.elance.com/>
 
 
4) point a DNS at this server only for those pages that rely
upon the LAMP
environment, but leave everything else "in
situ".  What would be the best
way to do that? Could www2.yoursite.com be made to point to
a different
server from www.yoursite.com <http://www.yoursite.com&g
t;  (the difference
being the insertion of the Arabic numberal two just after
the www, or is
that not workable?)
 
Apart from the obvious potential downwides (i.e. the guy who
bids the job
proves not be sufficiently capable) is there a fly in this
ointment?  Seems
like the upside would be to enable public libraries to
participate in some
of the innovation that academic libraries are able to access
by virtue of
being embedded in technically forward looking environments
with lots of
computer science folks running around pushing the envelope,
etc.
 
Regards,
Mark

      
 
 
_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

Aside from having a good firewall to sit behind, what is the
biggest reason against doing this in-house?

We just put together an Ubuntu linux server for about $600
with 1 GB of RAM and 400 GB of storage space.  

Ubuntu Dapper Drake Server with LAMP took an afternoon to
install.  Not *quite* out of the box solution, but pretty
darn close.  The configuration was pretty minimal, and I
just copied the web site over and it just worked.

We just had to get a DNS set up for that box (two actually,
one for the public, and one for a development version of the
web page), and holes poked in the firewall for HTTP and SSH,
and we went live with it inside of a week or two.

Now, we have our own environment in which to gradually
expand our capabilities.  I don't know how to do all the
things we want to do, but gradually, I find something
that's interesting, ask questions, read about it, and then
if it doesn't seem too hard to take on, I try it out.  We
don't have to ask permission to try anything new; we just
do all our playing in our development virtual host, and then
when it is ready to go live, we copy it over into
production.

I wouldn't call this a turnkey solution, but the
configuration was a lot less than what I was expecting, and
the support out there (and in here) is abundant.

It just seems to me that we're finally at a point where
this technology is affordable, the tools are (relatively)
easy to use, and the support is readily available.

Keith

Keith Engwall
Head of Library Systems and Technology
Catawba College Library
kengwallcatawba.edu
http://www.lib.catawba.edu



-----Original Message-----
From: web4lib-bounceswebjunction.org
[mailto:web4lib-bounceswebjunction.org] 
Sent: Tuesday, July 25, 2006 12:08 PM
To: 'Web4Lib (web4libwebjunction.org)'
Subject: [Web4lib] blue sky thinking

Hi,
 
This is just blue sky thinking and not pertaining to any
institution in particular.  What if an institution wanted to
gain the benefits that accrue to libraries that have
excellent server access and support in, say, a LAMP (Linux /
MySQL / Apache / Php & PERL) environment but had limited
means and possibilities.  Is there any reason why something
like this scenario wouldn't work?
 
1)  buy cheap space on a Linux shared hosting server that
provides root access such as 
      http://www.spry.com/pl
esk-vps/ <http://www.spry.co
m/plesk-vps/>  /
htt
p://support.jodohost.com/showthread.php?t=1726
<http://support.jodohost.com/showthread.php?t=1726> 
/ http://www
.linode.com/products/linodes.cfm
<http:/
/www.linode.com/products/linodes.cfm> 
 
2) identify open source tools that significantly expand
capabilities (i.e.
Content Management Systems or other database server driven
tools)
 
3)  either install and configure these tools oneself or
simply outsource it to someone via a site like http://www.elance.com/
<http://www.elance.com/>
 
 
4) point a DNS at this server only for those pages that rely
upon the LAMP environment, but leave everything else
"in situ".  What would be the best way to do
that? Could www2.yoursite.com be made to point to a
different server from www.yoursite.com <http://www.yoursite.com&g
t;  (the difference being the insertion of the Arabic
numberal two just after the www, or is that not workable?)
 
Apart from the obvious potential downwides (i.e. the guy who
bids the job proves not be sufficiently capable) is there a
fly in this ointment?  Seems like the upside would be to
enable public libraries to participate in some of the
innovation that academic libraries are able to access by
virtue of being embedded in technically forward looking
environments with lots of computer science folks running
around pushing the envelope, etc.
 
Regards,
Mark

      
 
 
_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

Aside from having a good firewall to sit behind, what is the
biggest reason against doing this in-house?

We just put together an Ubuntu linux server for about $600
with 1 GB of RAM and 400 GB of storage space.  

Ubuntu Dapper Drake Server with LAMP took an afternoon to
install.  Not *quite* out of the box solution, but pretty
darn close.  The configuration was pretty minimal, and I
just copied the web site over and it just worked.

We just had to get a DNS set up for that box (two actually,
one for the public, and one for a development version of the
web page), and holes poked in the firewall for HTTP and SSH,
and we went live with it inside of a week or two.

Now, we have our own environment in which to gradually
expand our capabilities.  I don't know how to do all the
things we want to do, but gradually, I find something
that's interesting, ask questions, read about it, and then
if it doesn't seem too hard to take on, I try it out.  We
don't have to ask permission to try anything new; we just
do all our playing in our development virtual host, and then
when it is ready to go live, we copy it over into
production.

I wouldn't call this a turnkey solution, but the
configuration was a lot less than what I was expecting, and
the support out there (and in here) is abundant.

It just seems to me that we're finally at a point where
this technology is affordable, the tools are (relatively)
easy to use, and the support is readily available.

Keith

Keith Engwall
Head of Library Systems and Technology
Catawba College Library
kengwallcatawba.edu
http://www.lib.catawba.edu



-----Original Message-----
From: web4lib-bounceswebjunction.org
[mailto:web4lib-bounceswebjunction.org] 
Sent: Tuesday, July 25, 2006 12:08 PM
To: 'Web4Lib (web4libwebjunction.org)'
Subject: [Web4lib] blue sky thinking

Hi,
 
This is just blue sky thinking and not pertaining to any
institution in particular.  What if an institution wanted to
gain the benefits that accrue to libraries that have
excellent server access and support in, say, a LAMP (Linux /
MySQL / Apache / Php & PERL) environment but had limited
means and possibilities.  Is there any reason why something
like this scenario wouldn't work?
 
1)  buy cheap space on a Linux shared hosting server that
provides root access such as 
      http://www.spry.com/pl
esk-vps/ <http://www.spry.co
m/plesk-vps/>  /
htt
p://support.jodohost.com/showthread.php?t=1726
<http://support.jodohost.com/showthread.php?t=1726> 
/ http://www
.linode.com/products/linodes.cfm
<http:/
/www.linode.com/products/linodes.cfm> 
 
2) identify open source tools that significantly expand
capabilities (i.e.
Content Management Systems or other database server driven
tools)
 
3)  either install and configure these tools oneself or
simply outsource it to someone via a site like http://www.elance.com/
<http://www.elance.com/>
 
 
4) point a DNS at this server only for those pages that rely
upon the LAMP environment, but leave everything else
"in situ".  What would be the best way to do
that? Could www2.yoursite.com be made to point to a
different server from www.yoursite.com <http://www.yoursite.com&g
t;  (the difference being the insertion of the Arabic
numberal two just after the www, or is that not workable?)
 
Apart from the obvious potential downwides (i.e. the guy who
bids the job proves not be sufficiently capable) is there a
fly in this ointment?  Seems like the upside would be to
enable public libraries to participate in some of the
innovation that academic libraries are able to access by
virtue of being embedded in technically forward looking
environments with lots of computer science folks running
around pushing the envelope, etc.
 
Regards,
Mark

      
 
 
_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

Keith D. Engwall wrote:
> Aside from having a good firewall to sit behind, what
is the biggest reason against doing this in-house?
>
> We just put together an Ubuntu linux server for about
$600 with 1 GB of RAM and 400 GB of storage space.  
>
>
>   
Having heard many statements/arguments for using open source
software, 
about a year ago I decided to do the experiment myself. I
meant to keep 
notes about how long things took me, what problems I ran
into, etc. but 
it was so frustrating that I just couldn't immortalize it
in a fixed 
form. But it went something like this:

1) Get operating system and burn to disk. Since I was
starting with a 
blank slate, I had to find a copy of the OS and get it onto
a CD so I 
could install it. The particular application I was aiming at
wanted that 
to be Debian. This was easy, maybe a half hour (note: I read
the install 
documentation before downloading most software).

2) Insert disk and boot up. Answer install questions. Here I
got stuck 
for an entire afternoon, believe it or not. First, there
were lots of 
questions to which I did not know the answer, so I was back
to my other 
machine to read more documentation. Then there was one menu
that 
absolutely stumped me -- not because I didn't know the
answer, but 
because I didn't know how to make an "x" appear
before my choices. I 
moved the cursor (tab tab tab) to the right spot and typed
"x" - 
nothing. I typed "a" "b" --
basically I went through the entire keyboard 
- nothing. Added "ctrl" before them - nothing.
Shift - nothing. I looked 
at the online documentation. It just said "select the
ones you wish to 
install." Nothing on how to select. The next day I
tried again, and only 
by accident did I hit the space bar - BINGO. So I wasted 4
hours because 
nowhere did it say that the way to select an item in a menu
is to use 
the space bar. OK. I got over that.

3) I now had a basic OS installed, but in order to run my
app I would 
need things like MySQL, Apache, etc. I would need
"packages." I have 
some experience with Red Hat (until they abandoned all of us
desktop 
users), with Suse, and with something else that I don't
remember now. 
Each has its own way to install software. I was on the phone
to a friend 
who is quite well versed in Unix, so he offered to help me
go through 
the package process. The first thing he told me to do was
type 
"dselect". De-select? To get things? No, it
means something like debian 
select, and you go through menus to get to a huge list of
possible 
software packages to select and install. All I can remember
about this 
is hours spent going through lists, selecting something I
needed, only 
to get a screen saying something about dependencies, but no
explanation 
of what I should DO about them. Many many hours. Eventually,
I had stuff 
installed, but no idea if any of it would work. I didn't
write it down, 
but this took days -- days because I would occasionally
install the 
wrong thing and then want to uninstall it, or I'd go to
install 
something and it would ask me questions I couldn't answer.
In the end, 
the thing that always frustrates me about Unix is that I
don't know 
WHERE my installed software is. There's a logic to it all,
I'm sure, but 
I've yet to find a clear explanation.

4) Now I installed the app I wanted to run, although it then
needed some 
changes to things like Apache, etc., to work properly. I
spent another 
day or so doing all of this. In the end, I actually was able
to start my 
app. But at that point I had run out of steam, and the
machine has been 
turned off since then. Maybe I'll spend my summer actually
getting the 
app up and running as a functioning system. Then again,
maybe not.

kc

-- 
-----------------------------------
Karen Coyle / Digital Library Consultant
kcoylekcoyle.net http://www.kcoyle.net
ph.: 510-540-7596
fx.: 510-848-3913
mo.: 510-435-8234
------------------------------------


_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

Keith D. Engwall wrote:
> Aside from having a good firewall to sit behind, what
is the biggest reason against doing this in-house?
>
> We just put together an Ubuntu linux server for about
$600 with 1 GB of RAM and 400 GB of storage space.  
>
>
>   
Having heard many statements/arguments for using open source
software, 
about a year ago I decided to do the experiment myself. I
meant to keep 
notes about how long things took me, what problems I ran
into, etc. but 
it was so frustrating that I just couldn't immortalize it
in a fixed 
form. But it went something like this:

1) Get operating system and burn to disk. Since I was
starting with a 
blank slate, I had to find a copy of the OS and get it onto
a CD so I 
could install it. The particular application I was aiming at
wanted that 
to be Debian. This was easy, maybe a half hour (note: I read
the install 
documentation before downloading most software).

2) Insert disk and boot up. Answer install questions. Here I
got stuck 
for an entire afternoon, believe it or not. First, there
were lots of 
questions to which I did not know the answer, so I was back
to my other 
machine to read more documentation. Then there was one menu
that 
absolutely stumped me -- not because I didn't know the
answer, but 
because I didn't know how to make an "x" appear
before my choices. I 
moved the cursor (tab tab tab) to the right spot and typed
"x" - 
nothing. I typed "a" "b" --
basically I went through the entire keyboard 
- nothing. Added "ctrl" before them - nothing.
Shift - nothing. I looked 
at the online documentation. It just said "select the
ones you wish to 
install." Nothing on how to select. The next day I
tried again, and only 
by accident did I hit the space bar - BINGO. So I wasted 4
hours because 
nowhere did it say that the way to select an item in a menu
is to use 
the space bar. OK. I got over that.

3) I now had a basic OS installed, but in order to run my
app I would 
need things like MySQL, Apache, etc. I would need
"packages." I have 
some experience with Red Hat (until they abandoned all of us
desktop 
users), with Suse, and with something else that I don't
remember now. 
Each has its own way to install software. I was on the phone
to a friend 
who is quite well versed in Unix, so he offered to help me
go through 
the package process. The first thing he told me to do was
type 
"dselect". De-select? To get things? No, it
means something like debian 
select, and you go through menus to get to a huge list of
possible 
software packages to select and install. All I can remember
about this 
is hours spent going through lists, selecting something I
needed, only 
to get a screen saying something about dependencies, but no
explanation 
of what I should DO about them. Many many hours. Eventually,
I had stuff 
installed, but no idea if any of it would work. I didn't
write it down, 
but this took days -- days because I would occasionally
install the 
wrong thing and then want to uninstall it, or I'd go to
install 
something and it would ask me questions I couldn't answer.
In the end, 
the thing that always frustrates me about Unix is that I
don't know 
WHERE my installed software is. There's a logic to it all,
I'm sure, but 
I've yet to find a clear explanation.

4) Now I installed the app I wanted to run, although it then
needed some 
changes to things like Apache, etc., to work properly. I
spent another 
day or so doing all of this. In the end, I actually was able
to start my 
app. But at that point I had run out of steam, and the
machine has been 
turned off since then. Maybe I'll spend my summer actually
getting the 
app up and running as a functioning system. Then again,
maybe not.

kc

-- 
-----------------------------------
Karen Coyle / Digital Library Consultant
kcoylekcoyle.net http://www.kcoyle.net
ph.: 510-540-7596
fx.: 510-848-3913
mo.: 510-435-8234
------------------------------------


_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


On Jul 27, 2006, at 10:25 AM, Karen Coyle wrote:

> Keith D. Engwall wrote:
>> Aside from having a good firewall to sit behind,
what is the  
>> biggest reason against doing this in-house?
>>
>> We just put together an Ubuntu linux server for
about $600 with 1  
>> GB of RAM and 400 GB of storage space.
>>
>>
> Having heard many statements/arguments for using open
source  
> software, about a year ago I decided to do the
experiment myself. I  
> meant to keep notes about how long things took me, what
problems I  
> ran into, etc. but it was so frustrating that I just
couldn't  
> immortalize it in a fixed form. But it went something
like this:
>
> 1) Get operating system and burn to disk. Since I was
starting with  
> a blank slate, I had to find a copy of the OS and get
it onto a CD  
> so I could install it. The particular application I was
aiming at  
> wanted that to be Debian. This was easy, maybe a half
hour (note: I  
> read the install documentation before downloading most
software).

There are many who would say you picked the "wrong
Linux Distro" to  
start but I applaud your efforts to stick with it. To this
well  
seasoned Debian user the installer is not particularly
"friendly".

http:/
/www.arouse.net/despair-linux/debian.jpg

I guess that's fair.

This comes up a lot in flamewars -- anyone that prefers the
command  
line to an ultra-shiny OS-X-ripoff-GUI is
"elitist".  That's simply  
not the case.  Those people ("we") started using
computers when there  
weren't shiny GUIs for Linux and we got used to the command
line.   
Since we have several years of experience using the command
line to  
do our work, we continue to do this and continue to
recommend it to  
others.  For us, it's faster, better, and cheaper and we
want to  
share that Goodness with you! (I'm using the command line
as an  
example... you can apply the same logic to emacs/LaTeX vs.  
OpenOffice, GMail vs. Mutt, Linux vs. Windows, etc.)

>
> 2) Insert disk and boot up. Answer install questions.
Here I got  
> stuck for an entire afternoon, believe it or not.
First, there were  
> lots of questions to which I did not know the answer,
so I was back  
> to my other machine to read more documentation. Then
there was one  
> menu that absolutely stumped me -- not because I
didn't know the  
> answer, but because I didn't know how to make an
"x" appear before  
> my choices. I moved the cursor (tab tab tab) to the
right spot and  
> typed "x" - nothing. I typed
"a" "b" -- basically I went through 

> the entire keyboard - nothing. Added "ctrl"
before them - nothing.  
> Shift - nothing. I looked at the online documentation.
It just said  
> "select the ones you wish to install."
Nothing on how to select.  
> The next day I tried again, and only by accident did I
hit the  
> space bar - BINGO. So I wasted 4 hours because nowhere
did it say  
> that the way to select an item in a menu is to use the
space bar.  
> OK. I got over that.

That is also a fair evaluation of Debian's
"Installer".

>
> 3) I now had a basic OS installed, but in order to run
my app I  
> would need things like MySQL, Apache, etc. I would need
"packages."  
> I have some experience with Red Hat (until they
abandoned all of us  
> desktop users), with Suse, and with something else that
I don't  
> remember now. Each has its own way to install software.
I was on  
> the phone to a friend who is quite well versed in Unix,
so he  
> offered to help me go through the package process. The
first thing  
> he told me to do was type "dselect".
De-select? To get things? No,  
> it means something like debian select, and you go
through menus to  
> get to a huge list of possible software packages to
select and  
> install. All I can remember about this is hours spent
going through  
> lists, selecting something I needed, only to get a
screen saying  
> something about dependencies, but no explanation of
what I should  
> DO about them. Many many hours. Eventually, I had stuff
installed,  
> but no idea if any of it would work. I didn't write it
down, but  
> this took days -- days because I would occasionally
install the  
> wrong thing and then want to uninstall it, or I'd go
to install  
> something and it would ask me questions I couldn't
answer. In the  
> end, the thing that always frustrates me about Unix is
that I don't  
> know WHERE my installed software is. There's a logic
to it all, I'm  
> sure, but I've yet to find a clear explanation.

I am REALLY surprised that the advice you got complicated
your life  
thusly. DSelect is okay (for me) on the rare occasion that
an  
upstream software simply won't work. For Apache/MySQL etc
apt-get  
would have been "my" pick.

"sudo apt-get install " would have
done ALL of this for  
you and automated the entire extremely frustrating
experience into a  
marvel.

Debian's packaging policy (I am picking Debian because I
have used it  
for a while) is very strict about where files are installed.
More on  
the QA procedures below

http://www.debian.org/doc/developers-reference
/ch-developer-duties


>
> 4) Now I installed the app I wanted to run, although it
then needed  
> some changes to things like Apache, etc., to work
properly. I spent  
> another day or so doing all of this. In the end, I
actually was  
> able to start my app. But at that point I had run out
of steam, and  
> the machine has been turned off since then. Maybe I'll
spend my  
> summer actually getting the app up and running as a
functioning  
> system. Then again, maybe not.

Unfortunately here also you are right. System Administration
tends to  
be a full time job. It is tedious, fascinating, challenging
and  
boring all at the same time. I feel like we have veered
sufficiently  
OT though.  I just try
to make sure Debian doesn't continue to be  
represented in unflattering ways. 

IMHO if you had picked apt-get to install your packages, you
probably  
would have spent all that energy with the custom
configuration. I  
digress...

regards,
./fxk
===============
Francis Kayiwa
Library Systems Team
4-180, MC 234
T: +1.312.996.2716
W: http://www.uic.edu/~kayiwa

Key: http://tigge
r.uic.edu/~kayiwa/kayiwa.gpg


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)

iD8DBQFEyOVGN+YGKSXdLhkRAhabAJ0SDQWKwU08fQ7x2QTKKhNdh7RrigCf
QNCR
Z5KYmXWBeEqomRCGeSn8Aik=
=WFgD
-----END PGP SIGNATURE-----
_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


On Jul 27, 2006, at 10:25 AM, Karen Coyle wrote:

> Keith D. Engwall wrote:
>> Aside from having a good firewall to sit behind,
what is the  
>> biggest reason against doing this in-house?
>>
>> We just put together an Ubuntu linux server for
about $600 with 1  
>> GB of RAM and 400 GB of storage space.
>>
>>
> Having heard many statements/arguments for using open
source  
> software, about a year ago I decided to do the
experiment myself. I  
> meant to keep notes about how long things took me, what
problems I  
> ran into, etc. but it was so frustrating that I just
couldn't  
> immortalize it in a fixed form. But it went something
like this:
>
> 1) Get operating system and burn to disk. Since I was
starting with  
> a blank slate, I had to find a copy of the OS and get
it onto a CD  
> so I could install it. The particular application I was
aiming at  
> wanted that to be Debian. This was easy, maybe a half
hour (note: I  
> read the install documentation before downloading most
software).

There are many who would say you picked the "wrong
Linux Distro" to  
start but I applaud your efforts to stick with it. To this
well  
seasoned Debian user the installer is not particularly
"friendly".

http:/
/www.arouse.net/despair-linux/debian.jpg

I guess that's fair.

This comes up a lot in flamewars -- anyone that prefers the
command  
line to an ultra-shiny OS-X-ripoff-GUI is
"elitist".  That's simply  
not the case.  Those people ("we") started using
computers when there  
weren't shiny GUIs for Linux and we got used to the command
line.   
Since we have several years of experience using the command
line to  
do our work, we continue to do this and continue to
recommend it to  
others.  For us, it's faster, better, and cheaper and we
want to  
share that Goodness with you! (I'm using the command line
as an  
example... you can apply the same logic to emacs/LaTeX vs.  
OpenOffice, GMail vs. Mutt, Linux vs. Windows, etc.)

>
> 2) Insert disk and boot up. Answer install questions.
Here I got  
> stuck for an entire afternoon, believe it or not.
First, there were  
> lots of questions to which I did not know the answer,
so I was back  
> to my other machine to read more documentation. Then
there was one  
> menu that absolutely stumped me -- not because I
didn't know the  
> answer, but because I didn't know how to make an
"x" appear before  
> my choices. I moved the cursor (tab tab tab) to the
right spot and  
> typed "x" - nothing. I typed
"a" "b" -- basically I went through 

> the entire keyboard - nothing. Added "ctrl"
before them - nothing.  
> Shift - nothing. I looked at the online documentation.
It just said  
> "select the ones you wish to install."
Nothing on how to select.  
> The next day I tried again, and only by accident did I
hit the  
> space bar - BINGO. So I wasted 4 hours because nowhere
did it say  
> that the way to select an item in a menu is to use the
space bar.  
> OK. I got over that.

That is also a fair evaluation of Debian's
"Installer".

>
> 3) I now had a basic OS installed, but in order to run
my app I  
> would need things like MySQL, Apache, etc. I would need
"packages."  
> I have some experience with Red Hat (until they
abandoned all of us  
> desktop users), with Suse, and with something else that
I don't  
> remember now. Each has its own way to install software.
I was on  
> the phone to a friend who is quite well versed in Unix,
so he  
> offered to help me go through the package process. The
first thing  
> he told me to do was type "dselect".
De-select? To get things? No,  
> it means something like debian select, and you go
through menus to  
> get to a huge list of possible software packages to
select and  
> install. All I can remember about this is hours spent
going through  
> lists, selecting something I needed, only to get a
screen saying  
> something about dependencies, but no explanation of
what I should  
> DO about them. Many many hours. Eventually, I had stuff
installed,  
> but no idea if any of it would work. I didn't write it
down, but  
> this took days -- days because I would occasionally
install the  
> wrong thing and then want to uninstall it, or I'd go
to install  
> something and it would ask me questions I couldn't
answer. In the  
> end, the thing that always frustrates me about Unix is
that I don't  
> know WHERE my installed software is. There's a logic
to it all, I'm  
> sure, but I've yet to find a clear explanation.

I am REALLY surprised that the advice you got complicated
your life  
thusly. DSelect is okay (for me) on the rare occasion that
an  
upstream software simply won't work. For Apache/MySQL etc
apt-get  
would have been "my" pick.

"sudo apt-get install " would have
done ALL of this for  
you and automated the entire extremely frustrating
experience into a  
marvel.

Debian's packaging policy (I am picking Debian because I
have used it  
for a while) is very strict about where files are installed.
More on  
the QA procedures below

http://www.debian.org/doc/developers-reference
/ch-developer-duties


>
> 4) Now I installed the app I wanted to run, although it
then needed  
> some changes to things like Apache, etc., to work
properly. I spent  
> another day or so doing all of this. In the end, I
actually was  
> able to start my app. But at that point I had run out
of steam, and  
> the machine has been turned off since then. Maybe I'll
spend my  
> summer actually getting the app up and running as a
functioning  
> system. Then again, maybe not.

Unfortunately here also you are right. System Administration
tends to  
be a full time job. It is tedious, fascinating, challenging
and  
boring all at the same time. I feel like we have veered
sufficiently  
OT though.  I just try
to make sure Debian doesn't continue to be  
represented in unflattering ways. 

IMHO if you had picked apt-get to install your packages, you
probably  
would have spent all that energy with the custom
configuration. I  
digress...

regards,
./fxk
===============
Francis Kayiwa
Library Systems Team
4-180, MC 234
T: +1.312.996.2716
W: http://www.uic.edu/~kayiwa

Key: http://tigge
r.uic.edu/~kayiwa/kayiwa.gpg


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)

iD8DBQFEyOVGN+YGKSXdLhkRAhabAJ0SDQWKwU08fQ7x2QTKKhNdh7RrigCf
QNCR
Z5KYmXWBeEqomRCGeSn8Aik=
=WFgD
-----END PGP SIGNATURE-----
_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

Karen: some thoughts and comments
(Warning: I'm a biased open source geek and 
command-line lover).

> Having heard many statements/arguments for using open
source software, about 
> a year ago I decided to do the experiment myself.

One initial question I ask is what closed-source software do
you use for 
these same things?  I can't tell what you're comparing the
various 
open-software applications to.  Windows Server?  Were you
setting up a 
desktop machine (I assume not since you talk about setting
up Apache)? 
What other servers have you set up with closed systems?

>
> 1) Get operating system and burn to disk. Since I was
starting with a blank 
> slate, I had to find a copy of the OS and get it onto a
CD so I could 
> install it. The particular application I was aiming at
wanted that to be 
> Debian. This was easy, maybe a half hour (note: I read
the install 
> documentation before downloading most software).

Debian is a pretty good linux distro, but it's well-known
for not being 
very "newbie" friendly for the standard install.
 Why did you choose that 
particular distribution?

>
> 2) Insert disk and boot up. Answer install questions.
Here I got stuck for 
> an entire afternoon, believe it or not. First, there
were lots of questions 
> to which I did not know the answer, so I was back to my
other machine to 
> read more documentation. Then there was one menu that
absolutely stumped me 
> -- not because I didn't know the answer, but because I
didn't know how to 
> make an "x" appear before my choices. I
moved the cursor (tab tab tab) to 
> the right spot and typed "x" - nothing. I
typed "a" "b" -- basically I went 
> through the entire keyboard - nothing. Added
"ctrl" before them - nothing. 
> Shift - nothing. I looked at the online documentation.
It just said "select 
> the ones you wish to install." Nothing on how to
select. The next day I 
> tried again, and only by accident did I hit the space
bar - BINGO. So I 
> wasted 4 hours because nowhere did it say that the way
to select an item in 
> a menu is to use the space bar. OK. I got over that.

Terminal-type applications can be annoying in this regard. 
Still, the 
typical buttons to try pressing are return and spacebar. 
Again, debian 
installer is probably one of the least "newbie"
friendly of all the 
distros.



> 3) I would need "packages."
<snip>long dialog

Packaging systems have always been a bit of a pain for
distributions. 
They're a nice idea, but sometimes it's just a headache. 
Again, I'm not 
sure what you're comparing it too.  Windows doesn't have
anything 
resembling a packaging system.  If you didn't want to deal
with the 
packages, why not download the source or binary of the
various 
applications?  That's what you would have had to do in
windows.

Ubuntu is based off of Debian and has a graphical interface.
  I also know 
that there are graphical front-ends for package management
for Debian.

Paths of packages tend to be one of the bad things in my
opinion about 
packaging systems.  Each distro tends to have it's own
logic.  Bad package 
paths is one of the factors that soured me on Redhat. I've
seen Windows 
apps install in wacky and bizarre places as well, or rely on
funky 
registry settings.

> 4) Now I installed the app I wanted to run, although it
then needed some 
> changes to things like Apache, etc., to work properly.
I spent another 
> day or so doing all of this. In the end, I actually was
able to start my 
> app.

You still need to configure any application, even commerical
ones.  Not 
sure where you're going with here.  Again, what are you
comparing this 
against?  Setting up a server properly is always tricky the
first time you 
do it for any system.  There's just so much customizing
that can be done.


Open source being, well, open is one of the major reasons I
like it.  I 
remember an old Windows game I had.  I recently tried to
re-install it on 
an upgraded system.  It checked to make sure I had a certain
version of a 
driver.  Problem was it required version 7, even though I
knew the later 
version 8 was backwards compatible.  I had to do some
bizarre workarounds 
to get it working in Windows.  In open source I could have
modified 
installer directly. I could have even rewritten the graphics
engine if I 
really felt like it


The cost of open source is that it will take time, or you
need to hire 
someone to take the time to do it.  An additional benefit is
you get the 
results of everyone else's time.  The newer versions of
Windows Server 
have convinced me that Windows is moving in a better
direction, so it's 
not as much of a matter of if I want something done I need
to use 
Linux/Unix.  But I still tend to use open source when I have
a choice 
between two equally good systems.

I'd recommend the next time you want to dip your toes into
open source to 
a) invest in some O'Reilly books.  (Running Linux, Apache,
any others for 
services you want to provide), b) talk with some people to
find good 
newbie distros, c) Don't run debian, d) don't run debian
;)



Jon Gorman

_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

Karen: some thoughts and comments
(Warning: I'm a biased open source geek and 
command-line lover).

> Having heard many statements/arguments for using open
source software, about 
> a year ago I decided to do the experiment myself.

One initial question I ask is what closed-source software do
you use for 
these same things?  I can't tell what you're comparing the
various 
open-software applications to.  Windows Server?  Were you
setting up a 
desktop machine (I assume not since you talk about setting
up Apache)? 
What other servers have you set up with closed systems?

>
> 1) Get operating system and burn to disk. Since I was
starting with a blank 
> slate, I had to find a copy of the OS and get it onto a
CD so I could 
> install it. The particular application I was aiming at
wanted that to be 
> Debian. This was easy, maybe a half hour (note: I read
the install 
> documentation before downloading most software).

Debian is a pretty good linux distro, but it's well-known
for not being 
very "newbie" friendly for the standard install.
 Why did you choose that 
particular distribution?

>
> 2) Insert disk and boot up. Answer install questions.
Here I got stuck for 
> an entire afternoon, believe it or not. First, there
were lots of questions 
> to which I did not know the answer, so I was back to my
other machine to 
> read more documentation. Then there was one menu that
absolutely stumped me 
> -- not because I didn't know the answer, but because I
didn't know how to 
> make an "x" appear before my choices. I
moved the cursor (tab tab tab) to 
> the right spot and typed "x" - nothing. I
typed "a" "b" -- basically I went 
> through the entire keyboard - nothing. Added
"ctrl" before them - nothing. 
> Shift - nothing. I looked at the online documentation.
It just said "select 
> the ones you wish to install." Nothing on how to
select. The next day I 
> tried again, and only by accident did I hit the space
bar - BINGO. So I 
> wasted 4 hours because nowhere did it say that the way
to select an item in 
> a menu is to use the space bar. OK. I got over that.

Terminal-type applications can be annoying in this regard. 
Still, the 
typical buttons to try pressing are return and spacebar. 
Again, debian 
installer is probably one of the least "newbie"
friendly of all the 
distros.



> 3) I would need "packages."
<snip>long dialog

Packaging systems have always been a bit of a pain for
distributions. 
They're a nice idea, but sometimes it's just a headache. 
Again, I'm not 
sure what you're comparing it too.  Windows doesn't have
anything 
resembling a packaging system.  If you didn't want to deal
with the 
packages, why not download the source or binary of the
various 
applications?  That's what you would have had to do in
windows.

Ubuntu is based off of Debian and has a graphical interface.
  I also know 
that there are graphical front-ends for package management
for Debian.

Paths of packages tend to be one of the bad things in my
opinion about 
packaging systems.  Each distro tends to have it's own
logic.  Bad package 
paths is one of the factors that soured me on Redhat. I've
seen Windows 
apps install in wacky and bizarre places as well, or rely on
funky 
registry settings.

> 4) Now I installed the app I wanted to run, although it
then needed some 
> changes to things like Apache, etc., to work properly.
I spent another 
> day or so doing all of this. In the end, I actually was
able to start my 
> app.

You still need to configure any application, even commerical
ones.  Not 
sure where you're going with here.  Again, what are you
comparing this 
against?  Setting up a server properly is always tricky the
first time you 
do it for any system.  There's just so much customizing
that can be done.


Open source being, well, open is one of the major reasons I
like it.  I 
remember an old Windows game I had.  I recently tried to
re-install it on 
an upgraded system.  It checked to make sure I had a certain
version of a 
driver.  Problem was it required version 7, even though I
knew the later 
version 8 was backwards compatible.  I had to do some
bizarre workarounds 
to get it working in Windows.  In open source I could have
modified 
installer directly. I could have even rewritten the graphics
engine if I 
really felt like it


The cost of open source is that it will take time, or you
need to hire 
someone to take the time to do it.  An additional benefit is
you get the 
results of everyone else's time.  The newer versions of
Windows Server 
have convinced me that Windows is moving in a better
direction, so it's 
not as much of a matter of if I want something done I need
to use 
Linux/Unix.  But I still tend to use open source when I have
a choice 
between two equally good systems.

I'd recommend the next time you want to dip your toes into
open source to 
a) invest in some O'Reilly books.  (Running Linux, Apache,
any others for 
services you want to provide), b) talk with some people to
find good 
newbie distros, c) Don't run debian, d) don't run debian
;)



Jon Gorman

_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/

> > 4) Now I installed the app I wanted to run,
although it then needed
> > some changes to things like Apache, etc., to work
properly. I spent
> > another day or so doing all of this. In the end, I
actually was
> > able to start my app. But at that point I had run
out of steam, and
> > the machine has been turned off since then. Maybe
I'll spend my
> > summer actually getting the app up and running as
a functioning
> > system. Then again, maybe not.
> 
> Unfortunately here also you are right. System
Administration tends to
> be a full time job. It is tedious, fascinating,
challenging and
> boring all at the same time. I feel like we have veered
sufficiently
> OT though.  


... That's not off-topic; that IS the topic. We have been
discussing the
pros and cons of maintaining in-house servers. That it is a
"full time job"
is a flag. At some point economies of scale kick in and
it's worthwhile to
maintain a local server (and for some services, and not for
others). But
even then, the operating system and software have to be
appropriately scaled
and supportable in-house, however that support system is
defined. 

Look at your resources, your goals, your obstacles, your
opportunities...
and, of course, your budget... I read one of Karen C's
points as being that
it's far more expensive than $600, something you
corroborate by saying the
human behind that server is a FT position, in IT no less. At
the loaded rate
(i.e. factoring in the overhead of a human body in an
organization plus
salary plus benefits), we're easily talking $100k per year.
That's
equivalent to one heckuva off-site server.

Karen G. Schneider
kgsbluehighways.com

_______________________________________________
Web4lib mailing list
Web4libwebjunction.org
http://lists.we
bjunction.org/web4lib/