hpr-knowledge-base/hpr_transcripts/hpr3413.txt

Episode: 3413
Title: HPR3413: Bash snippet - using coproc with SQLite
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr3413/hpr3413.mp3
Transcribed: 2025-10-24 22:58:27

---

This is Hacker Public Radio Episode 3413 for Wenzee, the first of September 2021.
Tid's show is entitled, Bash Snippet, using Coprock with Sklyte and is part of the series Bash,
scripting it is the 130th show of Dave Morris and is about 46 minutes long and carries an explicit flag.
The summary is, sending multiple queries to a running instance of Sklyte 3.
This episode of HPR is brought to you by archive.org.
Support universal access to all knowledge by heading over to archive.org forward slash donate.
Hello everybody, this is Dave Morris, welcome to Hacker Public Radio.
Today I'm talking about a Bash thing and it's one of the group shows I call Bash Snippets.
I'm talking about Coprock, the Command Coprock CoPROC and using it with the SQ Lite database thing.
I tell you where I'm coming from. I mean the process of rewriting some scripts that I use to
manage my collection of magnitude albums. Now I'm a lifetime magnitude member and I've talked
about this before and so I have access to the whole magnitude music collection which is a range
of music of various sorts but under magnitude as a sort of record label. I wrote a script for
downloading albums and putting them into my music directory way back. I talked about it in
2013 show 1204. Now the original scripts are available on GitLab still and in the lifetime
I've known one other person to make use of them. So I've been using them myself for quite a bit but
since 2013 I've written a few other support scripts. For example I wanted to be able to manage
a queue of albums that I wanted to buy and download and another one which summarizes the state
of this queue and it's this show queue script I'm currently updating. I included it in the resources
for the show calling it show underscore queue underscore orage or IG. The original version of this
script took magnitude album URLs from a file and this was my attempt at a queue. I've got some
things which take the top album off the queue and then edit it with said to just drop that off
and actually store that the way somewhere other as things that I've bought and in order to do this
to do the purchasing I needed to parse out but I can just use the URLs it stands for in order to
give me some sort of report. I need to parse out the last element of the URL which I fed through
GREP to get some pre-prepared summary information out of another file and the GitLab repository
shows this so I won't go into details here. Basically it's an XML database that
magnitude needs to produce and when I receive it I scan it and turn it into a file summer of
album summaries. We'll see this in a bit later on. There's not making any sense. Anyway,
my attitudes moved away from this XML file. XML's a difficult thing to manage and our
clutches recently done a show on XML. XML is great but people tend to use it as a sort of data store
as a database almost so people have used XML and found that it's incredibly easy to break the database
because XML is so fussy about the contents so you can't put less than and greater than in the actual
text. You can't put ampersands in. You have to jump through hoops a bit like HTML. You have to
jump through hoops in order to allow these through. Anyway magnitude moved from this XML file to an
SQL like database a few years ago now. I want to move my stuff toward to that. They're still
running the XML but it's not really supported so everything goes wrong with it then
I'm on my own so I want to move to their SQLite stuff and so for each URL that I find I want to
look it up in the database to find out its details. So the first version of this new script wasn't very
difficult to write what I had to do was to extract the search data from the URL so I'm reading
the URLs one at a time out of a file and parsing out the bit that's interesting and then using that
to do query on the database and just do that in in a loop and I've included my first attempt at this
as a file called show underscore q underscore db underscore one and it's one of the resources for
this episode so you can have a look at it if you want to. I felt bad about writing it this way because
the loop okay the loop is not going to iterate much at the moment it's got 26 things in the queue
because I'm not not downloading anything for a while so it's a moderately expensive process for
the loop to be calling SQLite every time around with a query with variable bits in it so I wasn't
happy about doing it that way and wanted to look for a better way. So in April of 2019 Clacker did a show
which is number 2793 and he talked about the bash coprock command so to summarize pretty much
what he said this command creates a subshell running a command or group of commands which is connected
to the calling or I tend to call it parent process I think it's not a invalid way of expressing it
the calling process through two file descriptors which I tend to call fd's in the notes here
it's possible for the calling shell to write to the input descriptor and read from the output one
and that's the way that you communicate with whatever is running in the subshell I was vaguely
aware of coprock at the time of Clacker's show but hadn't looked into it found the show fascinating
but I didn't have a use for the feature at the time so sort of went to the back of my mind really
so to solve my need to show my magnitude Q of future purchases it looked as if an SQLite
instance running in the subshell could be given queries one after another and return the answers
I needed so this was the sort of my journey to a bash script using coprock so let's go into some
details so the pending Q it just contains URLs and I've given an example of one it ends in a string
text this element is the so-called SKU which is I'd never come across this term until I saw
magnitude using but it's used all over the place it stands for stockkeeping units it's just a
a sort of index a unique ID for a thing you have in your stock collection in a warehouse or whatever
and that's the key to the album in the database or one of the keys anyway in the original XML
base system I when I run it I see the following example information when I run the current script
and the following stuff is a list of the artist the album the genres and the code the code being
the SKU so I get that for each item I'm going to show one here obviously so the original show Q
script just reads this this Q file and then it looks up the SKU in a file of reformatted XML
information as I've already said and I've also said that there's a copy of this script
available if you want to have a look at it you haven't included any of the files with it but just
to get the general principle of what's going on so talk a bit more about codeproc
like his show 2793 he talked about it about this command and how it behaves I looked at the
bash reference manual and there's a link at the end of the notes which documents it in a pretty
terse way I find this it's one of these things where you need I need to read it several times
for I start to get any idea of what it's talking about there's there's a lot more information in
there than a quick scan through will reveal I find so codeproc runs command as a subshell
which is generally called a code process and that's a generic term you see it quite a lot in the
the documentation and it's got this two-way pipe connected to it so you can talk to the
the codeproccess and the syntax which I think is a bit strange is the command codeproc
COPROC followed by an optional name followed by a command and then some redirections optional
the use of the name depends on the type of command if it's a simple command then no name can be
provided which I found confusing and I kept kept trying to give single commands a name in that
second sense you get the default the name is is allocated by the codeproc command itself and
it's just codeproc in capitals but because if that's the case you can only run one code process
out of time in your shell the alternatives are simple command as a command group as a complex
command group enclosed in braces or parentheses we've seen this in various bash things that we've
done I don't think I've actually concentrated much on what a compound command is but I've used them
so you probably if you've been following that you should understand what I'm talking about
now the user if you've sublime name under those circumstances with the compound command then
the you can't provide a name or not if you provide it then it's used obviously otherwise
COPROC in capitals is used so what's the name for well it's used to create variables relating
to the code process so the calling shell needs to have things to get in touch with the
the code process and what it does well first of all there is a variable created called
made from the name and then underscore pid in capitals that holds the process id number of the
sub shell of the code process and it also creates an array called the same as the name that you
provide the relevance of the array is that it contains the file descriptors for input and output
and element zero of your array contains the standard output from the code process and element one
has got the file descriptor for standard input now I know here in the in the the show notes that
I haven't really talked about file descriptors in the bash scripting shows that I've done but I'd
plan to do so before long talk about redirection as well that's really one of the next things
on the to-do list and yes okay COVID stopped me doing all this things and get messed up a lot of
people but my plan is to get back into doing some of these things because I found them quite fascinating
so don't mind digging around to learn more about them and it's also benefits me because I end up
knowing more with the consequence so here's an example of some simple usage and I've given
an example co-proc or where the command is the the bash date command there's not much point in
doing this the same the notes but just to get the idea of how it works so you would type at the
command line co-proc space date semicolon now don't don't hit return yet because you want to
see what comes back and the way that you do that is when the way that I'm doing it in this
example there are other ways is to use the cat command and then get till cat get its data from
and here's a redirection expression which which gets it from the element zero of the the array
that I was mentioning so you need to type less than which is a redirect set saying get the data
from here to cat and ampersand then in quotes double quotes dollar open brace co-proc in capitals
open square bracket zero close square bracket close brace close double quotes so what's that doing
it's the the dollar and then embraces name with an index you will remember that from when we
talked at length about arrays that's getting the contents of the array co-proc which is the default
because we're using that form of the command and it's the zero element and we're using that in a
redirect it actually just a number I don't know how these are allocated but it's usually 60 or
something like that then what else is grabbing file descriptor numbers on your system so you get
back from that a line which begins with square brackets and a number in my case it was one
the space and then a number 402 06 in my case that's the process ID number of the process that's
now being created for running the thing that you're running the next thing you get back is the date
the time that the co-processor ran which is 29th of july in this case and that's the result of what cat
did using that file descriptor and then you get another square bracketed one with a plus after it
and it's followed by the words done and then co-proc capital co-proc date so that's saying that the
process that you started is closed has been terminated and the reason that co-proc date and
the cat were both on the same line is because the co-proc the co-process created by co-proc will
fire up quickly do its stuff and then stop of course there's nothing to keep it going so if you
hit return at that point and then type the cat expression you'd be very lucky if you got the output
because the process would have gone doing it on the same line means that it's processed very
very quickly by the bash parser so you get the answer back I mean there's much easier ways of
finding out what the date is but it does give you some sort of idea of what this co-proc thing
can do I added a note to say if not in this particular case but if you are playing around
this and you found that your co-process doesn't stop well there it actually uses the the jobs
feature of bash which you haven't covered in the shows I've done you can type the jobs command
which shows any jobs that happen to be running using this mechanism it's usually things that have
been put into the background by various commands and if you did that in this particular case if you
could do that immediately after you'd fired off the co-proc then you would get back that there
was a job number one running that's why you get one in square brackets if you ever did end up
the hanging co-process you can issue the command kill space scared of l space and then percent one
that means kill job number one it uses these numbers as a simple means of accessing stuff you
can also kill the process number the PID process ID directly that's just in case you ever ever
do this and wonder what else going on if you bothered by the fact that co-proc generates this
square brackets number etc etc then you can suppress it if you're doing it on the command line
there are options that we'll do I'm not going to talk about them but when you run this stuff in
the script bash suppresses it anyway so mostly I don't care to be honest so let's look at the
gory details I've put it I've been spent a couple of weeks messing with this actually to try and
learn the intricacies of co-proc because I don't find it well documented and I don't fully
understand everything that's going on still haven't quite got to the bottom of it but I'm
got a lot further so using co-proc for single line stuff is pretty straightforward but there's
to be very little point in doing it to be honest to me anyway if you've got a co-proc that you
want to receive and send multiple lines of data then things get quite a bit more complicated
so I've listed three points that you need to be aware of so it's your co-processor you set
this up so you should have a good idea about whether it needs input it might I mean date didn't need
any input for example and all it produced was output you assume it's receiving what you send
by how it responds there's no easy way of telling whether anything you send to it gets there but
there's a whole issue about buffering you'll find that some bits of software will hold onto the
data that you sent them without actually actually processing it maybe because they're doing something
else perhaps or they might not be sending back stuff straight away so the buffering is an issue
second point when you finish sending stuff to the co-processor and you want to tell it's all done
you must you must close the input file descriptor and you do this with this rather arcane command
which consists of the bash command exec don't know if we've ever covered that I think we might have
done I've actually used it in things in functions I've I've talked about anyway exec is then followed
by in braces the name of the array that you you have created whatever that is I've just used name
as a generic marker here square brackets one so you're talking about the array element one which
is the input file descriptor and you follow that with a greater than sign an ampersand and a hyphen
see what I mean about arcane now this is one of the file descriptors slash redirection things
that bash offers and these I've literally left them till later on in the series of things I've
done for for bash about bash I've made reference to the the bash manual the section duplicating file
descriptors give some information about what this does doesn't really explain how you use the
operator mentioned there with a file descriptor held in an array and it certainly doesn't show this
odd form we're using here and explain it but I'll cover this later on in the bash tips
series and I've got my head around it and understood it fully the other thing is you can't really tell
how much output to read from the code process and you might be dealing with something that performs
IO buffering so hold on to its output unexpectedly and this can cause a deadlock if you get it wrong
and a deadlock is where a parent and a child process are both waiting for the other process
to do something so it's really easy to to get into a situation where you you for example run
cat against your coprop saying effectively give me what you've got and there's nothing so what do you
do you you've typed that at the command line or in your script and nothing's coming back you don't
then have a means of saying unless you go to another another terminal of saying um we'll just stop
then so it's messy to be saying so I've been playing with this and I wrote a coprop example script
which is included with the resources here called coprop underscore test dot sh and what it's
going to do is to run bash as a copropossess and it's going to just feed it bash commands
so my researchers came up with various things first one I declare a variable set a variable
process which is set to the string STD buff be with space hyphen i0 hyphen o0 hyphen e0
then a space then bash so what this does is it will run bash but with no buffering tells bash
not to do buffering at all and I found that this was going to work get this ticket example to work
then in the script I declare an indexed array called com whose contents are series of commands so
there's a date command who am i id an echo a print and so forth next is to set a variable in
to the number of elements in the array just like it add more and whatever without having to
to change a number we could practice anyway so the next thing is I start the copropossess the command
coprop space child i called it child and then in braces dollar process semicolon that's one of the
things about braces if you recall you can put each of the braces on separate lines with all of
the lines in the commands in between them on separate lines but if you don't you have to use
semicords and you have to have a semicolon before the last brace it that's ingrained into my head
now but there was a time when you used to really mess me up every time so we then set a variable i
to 0 because we're going to be counting and we go into a while loop the while loop has got a test
looking to see whether i the variable i is less than variable n so we're counting we're going to
count through the elements of an array and n is our last point at n minus one is the last point
so the first thing that happened in the loop is the command that we're going to get from the com
array is displayed with a dollar on the front just to make it easy to see which ones are the
commands and we send it to the code process when we do that by echoing it so it's echo double
quotes dollar open brace c o m open square bracket dollar i close square bracket close brace
close double quotes that gets you that the i th element of the array and we're redirecting that
into the file descriptor and that's achieved by typing and we saw that this before but it's not
quite the same greater than ampersand then we have an open double quotes dollar open brace
child open square brackets one close square bracket close brace closed double quotes so that
child one is the array element containing the input file descriptor so we're redirecting
output to that particular file descriptor and so we're sending a command to the code process
then not using cat this time because cat will just keep reading until it gets
handed file till it's run off the end of the data what i'm using here is read the read command
which will get just one line from whatever resource you're pointing in it whether it be a file or
whatever so the command is read space hyphen u this introduces the file descriptor that you're
going to read from by default you just read from standard in or whatever and we're a bit i
won't read the exact detail but in double quotes we've got the specification of the child array
element zero then we have hyphen r after space i do this by default because i'm told that it's
good practice to do it but i don't actually know the complete reason why follow the r with the
space and then the variable to receive the line and i call that results then echo that to the
to the standard out all the results then increment the variable i so that the loop can continue
so this will chug round sending the command to the code process receiving the output
now i added the the next bit just as an experiment to see how this would work
you'll have noticed i think that the the last command in that list of commands in the array is a
print f and it's printing the string hello world but there's a new line after the hello
another and after world so it's going to generate two lines and what we'll have seen now
for monitoring this this script is that it has written the first line of that from that print
but it hasn't done the second one because it was just the one read the the loop is now finished so
this bit of code says if then in double double square brackets hyphen v which is a thing that checks
whether a variable is exist isn't yeah i think whether it's some it's actually declared so if
child underscore capital pid exists let's say then that means that the process hasn't died
because we haven't told it to stop so it's still running one of the commands could have been
exit so we might have just terminated it that way so if it if the process is still running then
we echo two hyphons and the word end and two hyphons just to demarcate the point in the script
then we execute the thing that the command that closes the input stream and that is exec
open curly bracket open brace i should say child open square brackets one closed square brackets close
brace greater than ampersand hyphen and that's that arcane thing i talked about before which says
close this channel there's this file descriptor and then before leaving we run cat against
the inputs file descriptor from the process so that is cat and less than sign an ampersand
then in quotes this specification of child array element zero to get that file descriptor so
when you run that you see the output of each of the commands the date who am i id echo bash version
5.1.4 blah blah blah then the printf hello world produces hello then that if statement with the
end in it triggers the end and then the cat at the end gets the word world that's still in the pipe
waiting to be read so the loop just reads one line from the the co-process so that's what i was
talking about earlier on is that you don't necessarily know how many lines you're going to get back
from each iteration of the of the co-process each time you send it something you don't know how much
you're going to get back and there's no easy way of detecting where that is what the end is because
if you were to send a sort of end of end of file thing that would cause the file descriptor to be
closed which means you can't then talk to it anymore think i'm right in saying that so ask a
question in the summary of points here what would happen if a command was sent which produced no
output so if you added in quotes a colon to the list of commands a colon is just a
a null command in bash then the script would hang waiting for output that will never come so the
co-process would get the colon and that means do nothing so it will return no no data whatsoever
then the read would be triggered and it will hang their waiting for some to come back from the process
so the whole thing is frozen we're in we're in a sort of a deadlock now my experiments showed that
if i modified and i haven't actually demonstrated this in an in an example here if i modified the read
to include a timeout say five seconds or something then the read would fail at that point and would
then look carry on with the next command so that's one way of doing it but it's really clunky not
not very nice hopefully you'll have noticed that the output didn't include any of the job control
stuff that we saw before so there's no square bracketed ones and things so to my way of thinking
well this is a pretty useless example in the sense of doing anything of any consequence but it does
hopefully give some sort of indication of the way that co-process says behave and how you talk
to some stuff now having read the stuff about co-processes in ork back in the day when we were
doing a series on ork and deciding not to get into that one in the in the series i am nevertheless
thought i'd have a go at putting together something using co-processes in in
canoe ork you only works in canoe ork plain or it doesn't have this feature and so i've included
an example i won't talk about it but it's doing the same thing of being given an array of commands
and running bash and feeding it the commands and then reporting what comes back it's a little bit
nicer in some respects to my mind anyway but um i've included the script as part of the show
resources so have a look and see what you think so all of this led me to a solution to my
show cue script sort of solution anyway it it was an interesting voyage certainly understand
co-prog a bit better i don't think i'm going to use it but i should get a show out so i did
produce a final script which which works it does actually work and does do i want it to do but i'm
not sure i want to do it this way in the future it's included as part of the show as the resource
then it's called show underscore cue underscore db underscore two there's another one using the database
but a different way now i won't go into vast amount of detail here but it's listed in the notes
so you can have a look at it i tend to be i build my scripts on based around sort of template
so it tends to have a fair amount of standard preamble so there's a bunch of variables being defined
for where where files are the directories where they live and where the files are so i create
something that points to pending file which is the cue and another one that points to the latest
copy of the magnitude database which is sqlite underscore normalized dot db and there's some
sanity checks not enough actually you have to also check that the database existed check the cues
there but anyway then there's a check to see whether anything in the cue and there's an
appointment doing this thing going into further if the file is empty there's a regular expression
declared which contains the preamble of the magnitude URLs and the the final bit is the SKU stock
control thing almost that's fine that anyway that will come to that minute then i create a variable
called sql template i create it by using cat against a here document i'll cover here documents when
i get to redirection because it's regarded as piece of redirection it's just a way of getting
data from the script itself and plugging it into a file or a variable or whatever so what i'm
including here is piece of structured query language sql which is performing a select
on the database and because the way the database is designed is got a table for artists it's got a
table for albums it's got another one for genres another one for sub genres and it's also got
in each album it's got this code the SKU so in order to get what i want i need to join these tables
and do some jiggery polkery to get what i want this is not a database show so i won't go into
that much detail but the final bit of it or the penultimate bit of it is the line having SKU
equals and then open quote percent s closed quote that is because this whole string is a template
for use in a print f command and i want to substitute a value into that percent s when i
generate a version of it there's a semicolon at the end of that line because that's the end
of that piece of sql then the next line dot print is just a command to the sql sql
light three program there's it's got a sort of meta language which does all sorts of database
things but you can also get it to print stuff so i'm just getting it to print a line of hyphons
which you'll see in a minute when you see this being run so the codeproc is fired off next and i
called it dbproc i would use this std buff thing that i mentioned earlier on so it's switch buffering
off in the codeprocess and sql like to do buffering then sql like three gets the option
hyphen line which tells it to print it's the stuff out not in a table form but line by line
single line value and then dollar db is the path to the database itself so the main loop
we use a variable called n which is set to zero and we're reading in it's a wild loop and each
iteration we're reading a variable called url from somewhere semicolon do while read hyphen
r space url semicolon space do if you look at the end of that loop it's got a done and after done
is less than this redirection followed by in quotes the name q dollar q and what that does
is it makes the wild loop read the contents of the file and each read command that you issue
in the way that i've done gets the next line from there you see this before in other contexts
we increment this variable n and there's a test which compares the url that's returned
with that regular expression from earlier on so if they match it is a proper url then we should
get back a variable called bash rematch one which contains that last element which was in parentheses
in the regular expression we put that in a variable called sku if it doesn't match then there's a
problem with that url it's got mangled somehow so we put an error message showing which line it's
on that's what n is for and showing the url that's failing then last line really of the the loop
is a print f using that sku of template a new line on the end of it and we substitute in the
value of sku so that goes in as the percent s we saw earlier and then we pipe that to the process
using the input file script and i won't read that whole thing out there because we've done that
often already so the loop would just carry on sending multiple queries to the sickle light
process running as a code process so in this particular case once the loop is done we close off
the input stream which tells sickle light to all finished so sickle light will actually then send
it's probably what has already been sending data as it's going i could have written it too to pick
that up but um if i just go with with what i already had and um because i i'd learned that i could
do that as i was as i was proceeding with this so this this thing closes the the pipe but the
the code process output pipe will contain it will either contain all the output from sickle light
or sickle light will currently be waiting to flush what it has to something that's really going to
read it so we run cat to do that and we get back all of the lines that sickle light produced
though i've included the first two chunks first 12 lines of what comes back it's it's slightly
different from the original one because i'm getting the database to format the words
artist and the contents of artists so it's using sickle light's line mode which has to put out the
the title followed by an equals followed by the contents but yeah i can live with that and
there are two lines for the genre stuff there's a primary genre for our album and then some
supplementary genres so the first one is alt rock and it's also folk and instrumental new age
according to this um the xml version of the database didn't go to that depth of detail
obviously made the database a bit more detailed which is good so that's it i just wanted to sort of
tie the loose ends together uh with a conclusion from my point of view it's been an interesting
voyage take me a while to get what get there and uh it's been give me slight headaches
all the way i still don't understand the syntax of that exact and then a
ray ray element in braces with no dollar on the front i can't understand exactly what it is i
didn't know that was even legal but it apparently is i'd like to know where it's talking about it
hoping to find that along after this show goes out which one i'll tell you reading around lots of
other people's experiences co-proc one of the conclusions drawn is that named pipes are better
there is a facility in bash where you can create think called a named pipe which is a a device
which lets you write stuff and read stuff from different processes and talk to something that's
listening so you you can have process a sends the process b by writing through a name pipe and
b writes back to a through that the same name pipe it's a bidirectional thing hopefully i'll be
able to explain that better when i delve into a bit deeper in which case i'll talking about that
in the bash tips series a number of people said don't even do this use something like expect
expect as a program has been around since the early 90s i believe and it lets you run a sub
process of a thing usually it's a sort of like a co-proc would be it's something which runs
it's a command line thing that you can talk to by typing it at it from the command line
but it lets you run it as a sub process and write a program around it which lets
which which which runs it as if you were running it now i spent several years using expect
and expect tk in my job because we had a series of machines running altrix and they called
decafina on top of that where all of the the service level stuff did not have an api it just had
command line stuff so if you had you know like 500 students that you wanted to give account on
this system then you'd have to sit down for hours and hours and hours on end and type the stuff
in because the commands did not have a means of receiving data in a meaningful way so i
learnt expect expect tk more take the tk bit is the runs the tcl tk front end on top of expect
expect is written in tcl it offers tcl interface so that was great i couldn't i was able to
automate a lot of stuff using that i'm not used it for ten twenty years i don't know but anyway
expect is a great answer to this type of problem and if you go and look up the Wikipedia page
i've referred to here then you can see that expect itself is quite quite clever but there's quite
lot of derivatives of it versions written in various languages and there's also one which runs
as a bash command which is quite nice to to work with i'm just in the process of playing with it
and i'm quite enjoying it the other point that was made was have a look at the stack exchange
reference i've added here which where somebody breaks down co proc in enormous amount of detail
which is in relation to bash and other other shells so i guess my final point is that i can write
for this application i can write a simpler pulse script that connects the execute like database
prepares a query with a substitution point repeats the core different values without and i can
do that without a code process and i've added i did it just to prove that i could i probably use
that in fact and i've included it as a resource this is an unfinished fairly rough program but
just to give you some idea and loads of other programming solutions are available obviously
either nor c if you want c plus plus there are many many things rust i don't know i doesn't feel
to me as if this is a task for bash pushing up against the limits and bash not really catering
for it all that well and i think the conclusion i drew from clack his experiences with it was
that it's a feature looking for a use so his challenge was see if you can find a useful let me know
i thought i had and it's sort of a use but it's not ideal if you disagree then come back and
tell me why and that would be an interesting discussion and i preferably has a show of course
there's a bunch of examples here for you to look at should the mood take you and that's otherwise
everything okay i hope that wasn't too tedious and you got something out of it all right then bye bye
you've been listening to hecka public radio at hecka public radio dot org
we are a community podcast network that releases shows every weekday Monday through Friday
today's show like all our shows was contributed by an hbr listener like yourself
if you ever thought of recording a podcast then click on our contributing to find out how easy it
really is hecka public radio was founded by the digital dog pound and the infonomican computer club
and it's part of the binary revolution at binrev.com if you have comments on today's show
please email the host directly leave a comment on the website or record a follow up episode
yourself unless otherwise status today's show is released on the creative comments
attribution share a light 3.0 license