Hi,
I'm in trouble trying to make this:

A patch writes to text.txt a name of file.wav and a number of how many times it was played;
On a [loadbang] a patch reads file text.txt and knows how many times file.wav has been played and counts further;

I need to put more names of files with their times they been played in the same text.txt:
Times Name
12 file
3 file1
23 file2

How to define a line for each sub patch that it could read and write only in it's line. My patch gathering statistics of hundreds of files and would be nice to have it in one file.

I attached a patch with two sub patches for you to try to separate lines of sub patches that they could write and get different data from the text file according to the line they work on. Maybe you'll have other suggestions or even comments about it or modifications on my example patch.

more power together

http://www.pdpatchrepo.info/hurleur/Define_line.zip

Found this, but can't clearly understand

http://lists.puredata.info/pipermail/pd-list/2009-04/069735.html
http://lists.puredata.info/pipermail/pd-list/2009-04/069745.html
http://lists.puredata.info/pipermail/pd-list/2009-04/069734.html

I got a replay from

"Derek Holzer Pd-list
I recommend the [col] external for this, both writing to and reading from specific lines in a file."

Any ideas about what library contain a [col] GUI

i'm pretty sure, he meant [coll] from cyclone. but never worked with it. but it has a help file.
if it's not what you were looking for, i would think that it was maybe possible to load the file on startup into a list and then work with the list while the patch is running. before closing the patch just save the list again to a file.
here you can find a good overview for list objects and abstractions available in extended:
http://en.flossmanuals.net/PureData/ListofObjects

You could do this with [coll], but if you want to use an object that doesn't stray too far from [textfile], you could use [msgfile].

I changed my mind on how to sort the problem and I found that using of [coll], [msgfile] is a thing I can't work out now. I decided to crate separate files.txt of each audio file with a name and changing argument $1or %d (time of plays) in that name. Correct me if I'm not right, about the argument...

When I'll list ($ ls) a folder content in terminal window I'll see a clear listed statistics with names and number of plays as files but even can copy it as text list. Of course a number of play times will be in a file too for [qlist] to read it.

Sounds a little different than I imagine, but .... would be nice to do something like already prepared list of names in .txt, and then a sound plays reads a list and then found its name write a number at the end of the line.

Thank you.

Have you looked at Pd Container in Pd-extended? I think the object is [h_map]. It uses key->values for storing data

I haven't investigated it properly, but maybe this what you need?

Hello,
for me to understand and make it, the most helpful practice was to study how numset.pd file works for synth in help browser/stuff/synth. Its about when you add this abstraction in your window as an object giving name [numset name x]. No matter how many of unique names you have, they'll appear in one (or many) preset.txt file as:

list name-in <value>
list name2-in <value>

...enjoying pd, help files and growing experience