CD burning prog for RH9?

[mallowman]

Originally posted by leech
Just type 'apt-get install gcombust' oh wait, you're using RH... Ok, enough of me being an ass. [...] You would think that it would through the GUI, but apparently RH hasn't gotten the idea in their heads yet that MAYBE some of their users might actually want to be able to install 3rd party packages.

Actually, rawhide (RedHat's development snapshots) includes yum, and presumably it will be included in the next release (which if I remember correctly will be coming late September/early October) and hopefully in the second beta for that release. Although I have not tried yum yet, I am very interested in it since I found out from watching list traffic that it uses the python bindings for librpm so it should be quite a bit faster than apt4rpm which I read access the package database by calling rpm as a subprocess.

[Kooldino]

Originally posted by mallowman
What if you try installing the RPM from the command line? Use rpm -Uhv gcombust-0.1.55-1.fr.i386.rpm

Does it give any error messages? It might be that there is an unmet dependency that is keeping it from being installed.

rant: Why is it that for half the commands in linux, if I do a "command > blah.txt", it ignores it and outputs to the screen??


Ok, so anyway, I ran it on the command line...looks like I'm in RPM hell...

Failed dependencies:
cdda2wav >=1.1- is needed
cdlabelgen >=1.5.0 is needed

[Kooldino]

Originally posted by leech
Just type 'apt-get install gcombust' oh wait, you're using RH... Ok, enough of me being an ass.

What distrib does that? ANd what does apt-get do? Get all the dependencies?

Mallowman has it right, run it from the command line, it'll tell you if you are missing any dependancies. You would think that it would through the GUI, but apparently RH hasn't gotten the idea in their heads yet that MAYBE some of their users might actually want to be able to install 3rd party packages.

Yeah, seriously.

Edit: Indeed gcombust is in the freshrpms repository http://freshrpms.net. gcombust-0.1.55-1.fr.i386.rpm is the very filename of it. If you follow the instructions there and install apt4rpm then you can 'apt-get install gcombust' from the command line and it'll fetch it. While you're at it, I'd recommend installing synaptic as it is an excellent package installer/downloader for apt.

What's all this apt business?

[mallowman]

Originally posted by Kooldino
rant: Why is it that for half the commands in linux, if I do a "command > blah.txt", it ignores it and outputs to the screen??

This is because there are actually two output streams. One is called standard output (assigned the number 1) and the other is called standard error (assigned to number 2), which is where many programs will print diagnostic information. '<span style="font-family: fixed">&gt; blah.txt</span>' only redirects the standard output stream, to also redirect standard error, you must write something like this '<span style="font-family: fixed">&gt; blah.txt 2&gt;&amp;1</span>' which can be read as redirect standard output to blah.txt and then redirect standard error to where ever standard out is going. It is important not to switch it around and write '<span style="font-family: fixed">2&gt;&amp;1 &gt; blah.txt</span>' because that says redirect standard error to where ever standard out is going (the terminal) and then redirect standard out to blah.txt, so you still end up with standard error going to the screen. It is kind of confusing, it took me a little while to fully understand it, so I hope I explained it well enough.

Ok, so anyway, I ran it on the command line...looks like I'm in RPM hell...

Failed dependencies:
cdda2wav >=1.1- is needed
cdlabelgen >=1.5.0 is needed

These two packages are part of RH9 so you either need to install them from the installation media or download them from your favorite RedHat mirror like <a href="ftp://ftp.dulug.duke.edu/ftp.redhat.com/pub/redhat/linux/9/en/os/i386/RedHat/RPMS">this one</a>. Once you get those installed the gcombust package ought to install without a hitch.

[Kooldino]

Originally posted by mallowman
This is because there are actually two output streams. One is called standard output (assigned the number 1) and the other is called standard error (assigned to number 2), which is where many programs will print diagnostic information. '<span style="font-family: fixed">&gt; blah.txt</span>' only redirects the standard output stream, to also redirect standard error, you must write something like this '<span style="font-family: fixed">&gt; blah.txt 2&gt;&amp;1</span>' which can be read as redirect standard output to blah.txt and then redirect standard error to where ever standard out is going. It is important not to switch it around and write '<span style="font-family: fixed">2&gt;&amp;1 &gt; blah.txt</span>' because that says redirect standard error to where ever standard out is going (the terminal) and then redirect standard out to blah.txt, so you still end up with standard error going to the screen. It is kind of confusing, it took me a little while to fully understand it, so I hope I explained it well enough.

Sweet, that works, but WHY? I what does the "2>&1" transalte to exactly? I just don't follow the notation.

These two packages are part of RH9 so you either need to install them from the installation media or download them from your favorite RedHat mirror like <a href="ftp://ftp.dulug.duke.edu/ftp.redhat.com/pub/redhat/linux/9/en/os/i386/RedHat/RPMS">this one</a>. Once you get those installed the gcombust package ought to install without a hitch.

I'll give it a go. Hey, anyone know that rpm finder site offhand?

[mallowman]

Originally posted by Kooldino
Sweet, that works, but WHY? I what does the "2>&1" transalte to exactly? I just don't follow the notation.

Okay here's a more general explanation of redirection, at least as it occurs when using bash, other shells may vary. Sorry it is so long winded, it may be a bit (ha!) more detailed than it needs to be. Actually I think I will split it up into four different posts (nothing like a little post whoring 'eh?) to be a little more manageable. This one is the introduction, the second will be about output redirection, the third about input redirection, and the fourth about piping.

Here goes...

In Unix (and, therefore, Linux too) when you execute a program it inherits all of the open files from the process that started it. This allows the shell (the program you type commands into) to do some interesting things collectively known as redirection and pipes.

The basic symbols for redirection are '&gt;' for output redirection and '&lt;' for input redirection, sometimes these symbols are read as "chunks" but I'm not sure why. The symbol for piping is '|', the vertical bar, read outloud as "piped to". Each of these symbols can have a number in front of it which specifies which file number you are affecting.

If you do not specify a number, it defaults to 1 (standard output) for output redirection and 0 (standard input) for input redirection. The only other "standard" file number is standard error which is number 2.

At first it may seem needlessly complex to have two separate output streams, but it is very useful when executing programs in the background with no terminal attached to them. A script that run non interactively can execute a program which may normally produce a lot of output, but you would like to keep track of only the error messages. Or you need to parse the output of a program, but you don't want any error messages in the way. Neither would be impossible if there were not a separate file number for error messages. Of course some programs misbehave and put all of their output (errors and normal output) to the standard output file, so you need to find another strategy for those sorts of programs.

I'll give it a go. Hey, anyone know that rpm finder site offhand?

And way down here, hiding off by itself is the answer to the probably more pressing query: Are you thinking of <a href="http://www.rpmfind.net/">RPM Find</a>?

[mallowman]

output redirection

Output redirection is probably the most commonly encountered since many programs have a builtin option (usually named <span style="font-family: monospace">-f</span>) to use a file for their input. There are a few different kinds of output redirection, remember each one can have a number in front of it that says which file number is redirected.

<dl>
<dt style="font-family: monospace">&gt; <i>filename</i></dt>
<dd>This is regular output redirection, it sends everything written to the specified file number (defaults to standard output) to the specified file. If the file exists it is replaced. Some shells have an option (in bash it is called '<span style="font-family: monospace">noclobber</span>', type "<span style="font-family: monospace">help set</span>" for information on setting the option) that gives an error if the file exists instead of overwriting it.</dd>

<dt style="font-family: monospace">&gt;&gt; <i>filename</i></dt>
<dd>This is the same as regular redirection except it appends to the specified file instead of overwriting it. You would use this if you wanted to make a log file or wanted to collect the output of several commands in the same file.</dd>

<dt style="font-family: monospace">&gt;&amp;<i>number</i></dt>
<dd>This is known as duplicating redirection. It is a somewhat confusing name, but it reflects the underlying operating system function used to implement the feature. It says whatever location the number after the <span style="font-family: monospace">&gt;&amp;</span> is going to at that time, make the number before the <span style="font-family: monospace">&gt;&amp;</span> (defaulting to stadard output if not specified) go to that place too. It is important to note that the redirections are read right to left, so any redirections to the right of this one have not taken effect yet and is why "<span style="font-family: monospace">2&gt;&amp;1 &gt; file.txt</span>" will not change where standard error (file number 2) goes.</dd>

<dt style="font-family: monospace">&amp;&gt; <i>filename</i></dt>
<dd>I believe this form is redirection is a bash extension so other shells may not support it. It is a shortcut for a common use of the previous form (which is why it looks similar) which is to redirect standard output and standard error at the same time. Both standard output and standard error will be redirected to the specified file name. It has the same effect as "<span style="font-family: monospace">&gt; file.txt 2&gt;&amp;1</span>". This form does not accept a number before it to specifiy the file number to change, it always changes numbers 1 and 2.
</dd>
</dl>

Edit: fix the monospacing

[mallowman]

input redirection

Simple input redirection and duplicating input redirection works in a way that is pretty orthoganal to output redirection so there are no real surprises. The analogue to appending to output is a total different matter however. Here is a description of the input redirection, I will try to be more brief than with output redirection.

<dl>
<dt style="font-family: monospace">&lt; <i>filename</i></dt>
<dd>This assigns the given filename to the number specified before the redirection (defaulting to standard input, number 0, if not specified). Most programs will only look at standard input, so specifiying a file number isn't that useful.</dd>

<dt style="font-family: monospace">&lt;&amp;<i>number</i></dt>
<dd>This is input duplication, it assigns the filenumber before the redirection to be the same as the file number after the redirection. It has the same gotchas as output redirection. It would be used mainly in scripts, since usually only the standard descriptors are used by other programs. If for example the script was called with "<span style="font-family: monospace">script.sh 3&lt; somefile 4&lt; otherfile</span>", inside the script you could write "<span style="font-family: monospace">program &lt;&amp;3</span>" to send <span style="font-family: monospace">somefile</span> to the standard input of <span style="font-family: monospace">program</span> and "<span style="font-family: monospace">other &lt;&amp;4</span>" to send <span style="font-family: monospace">otherfile</span> to the standard input of <span style="font-family: monospace">other</span></dd>

<dt style="font-family: monospace">&lt;&lt;<i>word</i></dt>
<dd>This is a very special form of redirection known as the here-document. It takes everything typed after pressing enter until the specified word is by its self at the beginning of a line, and redirects it as the standard input of the program. For example:

<code style="white-space: pre">
foo &lt;&lt; EOF
this is the input to foo
you can also put this in script files just as I have typed here
EOF
</code>

It has the added feature of expanding any shell variables like "<span style="font-family: monospace">$HOME</span>" and command substitution like "<span style="font-family: monospace">`whoami`</span>" before sending the input to the program. There is a variant that puts a "<span style="font-family: monospace">-</span>" before the word that will ignore tab characters in the front of the lines of the here-documents. This lets you indent things nicely in scripts like this:

<code style="white-space: pre">
foo &lt;&lt;-EOF
this is an indented here-document.
it makes scripts more readable than the previous example
EOF
</code>
</dd>
</dl>

[mallowman]

pipes

The last kind of redirection, although it could really be put in a category all its own is piping. This is one of the things that makes the Unix command line so powerful and flexible. If you ever meet a Unix wizard, he will probably write really long command lines with many pipes that it will take somone else at least a few days to figure out. The basic concept of pipes is to send the output of one program into the input of another program. Unix (and Linux too) has tons of little programs (sometimes called filters) that take some input, modify it in someway, and then output the result. Some examples are cut, sort, grep, col, column, fmt, head, tail, fold, uniq, tr, sed, and many more. The GNU textutils package contains a lot of these filters (type "<span style="font-family: monospace">info textutils</span>" to get some documentation or look on the <a href="http://www.lepp.cornell.edu/public/COMP/info/textutils/textutils_toc.html">web</a>, especially check out the <b>Opening the software toolbox</b> section, it explains some of the culture behind this sort of design.

The basic syntax for piping is as follows:

<span style="font-family: monospace">command1 arguments | command2 arguments | command3 arguments <i>redirection</i></span>

Here are some example pipelines that I like to use a lot:

<code>
command | less <i># the ubiquitous pipe to a pager, lets me scroll through the output of a command, add it to the end of any of these if the output is too long</i>

du -kS | sort -rn <i># print out the size of each directory and sort according to size, the biggest directory appears at the bottom, useful to find out which directories to attack to free up som disk space, remove the S from the du options and it will include the size of subdirectories</i>

cat somefile | sort | uniq <i># show somefile with all of the duplicate lines removed. Uniq has lots of interesting options that change its output format, for instnace you can find out how many times each line was duplicated. Uniq only works on sorted files though, hence the sort just before uniq</i>

cat somefile | sed -e 's,searchstring,replacement,' <i># replace all occurences of searchstring with replacement, actually searchstring is a regular expression and replacement allows backreferences, so you can do some pretty powerfull stuff</i>

rpm -qa | grep searchstring <i># print a list of all packages that match the regular expression searchstring</i>

command 2&gt;&amp;1 | tee somefile <i># combines standard out and standard error showing them on the screen and also saving them in somefile at the same time. Tee is like a T joint in a water pipeline</i>
</code>

Actually anyplace I wrote command, you can put a whole pipeline of commands so you could tack on <span style="font-family: monospace">| tee somefile</span> onto any of the above and save the output into a file as well as seeing it on your screen.

Once you learn how powerful these constructs can be (along with simple shell scripting, tab completion, and a few other things) you can do some pretty amazing things with the command line. It will also make you curse everytime you are trying to do some sort of batch file operation on a windows computer, command.com is very limitied and cmd.exe is only slightly better, and neither has all of the various filter programs that are standard fare for unix. Eventually you get fed up and install <a href="http://www.cygwin.com">Cygwin</a> on every windows box you have to use.

[Kooldino]

Originally posted by mallowman
(post 1)

Good to go so far. And yeah, that was the right link.

[Kooldino]

Wow, thanks for the lesson, mallowman!

[leech]

Originally posted by Kooldino
What distrib does that? ANd what does apt-get do? Get all the dependencies?

What's all this apt business?

Any Debian based distro comes with Apt as part of it. (Apt = Automatic Package Tool) And yes it gets all the dependencies.

root@skint:/home/leech# apt-get install zinf
Reading Package Lists... Done
Building Dependency Tree... Done
The following extra packages will be installed:
libmusicbrainz2
Suggested packages:
zinf-plugin-esound zinf-plugin-arts
The following NEW packages will be installed:
zinf
1 packages upgraded, 1 newly installed, 0 to remove and 36 not upgraded.
Need to get 1588kB of archives.
After unpacking 3138kB of additional disk space will be used.
Do you want to continue? [Y/n]
Get:1 ftp://ftp.debian.org sid/main libmusicbrainz2 2.0.2-1 [102kB]
Get:2 ftp://ftp.debian.org sid/main zinf 2.2.4-4 [1486kB]
Fetched 1588kB in 24s (65.0kB/s)
(Reading database ... 167700 files and directories currently installed.)
Preparing to replace libmusicbrainz2 2.0.1-1 (using .../libmusicbrainz2_2.0.2-1_i386.deb) ...
Unpacking replacement libmusicbrainz2 ...
Selecting previously deselected package zinf.
Unpacking zinf (from .../archives/zinf_2.2.4-4_i386.deb) ...
Setting up libmusicbrainz2 (2.0.2-1) ...

Setting up zinf (2.2.4-4) ...

root@skint:/home/leech#


As a quick example Currently some of the different Distros that are debian based; Libranet (which is what I use), Xandros, Lindows, etc. Here is a link to omre.

http://www.debian.org/misc/children-distros

[TnT]

Another auto-dependency getting system is Portage in Gentoo. It's really easy to use, but I've never used Debian's apt-get, so I can't compare.

Thien

[leech]

I'm quite interested in everything that I've heard of Gentoo. Sounds pretty sweet. Though Debian has basically the same concept (even though sometimes it doesn't work as planned) called apt-build. It'll build whatever src package for optimizations.

With Debian you have Synaptic which is a GUI frontend for apt. And the newest version (.42) rocks!

http://fnord.dynu.com/Screenshot.png

As you can see, it shows Package Name, Installed Version, Available Version, Installed Size, and Description.

On the tabs below you have Information, Description (this will give a fuller description), Dependencies, installed Files.

On the dependencies tab you'll see some that are red, which means they aren't installed. It'll also let you select Suggests, and which packages depend on it. Pretty sweet, if I do say so myself. They do have Synaptic for RPM as well. (Originally it was part of the Connectiva Distro, which are the guys who first ported apt to rpm.)

Leech

[Kooldino]

leech - that's pretty badass. I like your desktop too.