Bug#927996: RFS: diskfit/2.0.2.3 [ITP] -- Simple disk fit calculator
Good morning,
an updated package is available at
https://launchpad.net/~velnias/+archive/ubuntu/sandbox/+sourcefiles/diskfit/
2.0.2.13-1~eoan~ppa1/diskfit_2.0.2.13-1~eoan~ppa1.dsc
Am Dienstag, 11. Juni 2019, 00:41:15 CEST schrieb Adam Borowski:
> Hi!
>
> On Wed, May 29, 2019 at 08:30:50AM +0200, Heiko Schäfer wrote:
>
> You'll need to update that for the final upload, but during the review
> that's ok.
Of course I will do that as soon as I have a final place to upload it.
> debian/copyright needs an entry for install-sh which is under the MIT/X11
> license.
I hope I got the right license text.
> debians/source/options forces xz level 9 compression, which for this size of
> files is harmful (doesn't improve disk space, and requires a lot of memory
> on tiny machines). Please just drop these settings -- they should be used
> only in special cases.
I've removed this file completely.
> debian/control: could you please fill in Homepage/Vcs-*, or at least drop
> the commented out fields?
done.
> The long desc really needs to be extended. Even with trying to use the
> program, I had some trouble finding out what it is for.
To describe it is a big challenge for me. I hope is is a bit clearer now.
I'm aware it is a quite special tool, but my search for an existing one I
could integrate in my daily workflow has been unsuccessful, so I decided to
write one for myself.
Over 3 years being a lot of help, I decided to make it available to a broader
audience by requesting it to get added to Debian.
> Also, a man page is vital for a command-line tool.
I added one for 'diskfit'. I did none for the GUI.
> Upstreamish side: I played with the GUI a bit, and it seems to handle
> selecting a dir badly (which, in my naive understanding, is what most
> people will try to do). It does nothing until you click "Cancel", in
> which case it'll add one of previously (not currently) show directories.
> And then the program doesn't seem to do anything useful. So, if dirs
> are not supported, the programs shouldn't at least try them.
First of all a *big* thank you for this extra testing!
At least in my version of Qt (5.12.3) it is - as wanted - not possible to just
select a directory. I guess you mean the contents of an entire directory.
Due to the mathematical nature of combinatorics the larger the amount of files
to process, the 'much' longer it takes. On my PC it takes around 7.5 h to
process 38 files. Good for an overnight calculation :-)
The progress bar is updated discretely at 1 % steps, so - if you have a lot of
input files - it can take a *long* time until you will see any progress.
This is unavoidable, because a non-discrete update (of the underlying) diskfit
process would slow down everything up to an absolutely unacceptable amount of
time. There is a second drawback, if you are patient and you're close to 50 %
it can happen, that the remaining 50 % will run nearly as fast as light.
The progress calculation is divided into two parts:
1) the calculation of all combinations (lengthy)
2) preparing the output (*can* be lengthy, but hard to predict)
The goal of this tool is to approach the target size as exactly as possible
and not to use heuristical algorithms and the most common use cases will do
that in a reasonable time. Of course it is not meant to recursively create a
set of optimally fitted DVDs for a huge collection of movies.
You could try out a similar example I use to generate the profile information
while building:
- choose the file pattern '[cdmMR]*' from src/diskfit, no directories just
files
- choose a custom target of 72240 bytes
This should give you two results and will run under one second.
If you have ideas how to communicate this behaviour to the user, I would be
thankful.
>
> Meow!
P.S. I love funeral doom too :-D
Reply to: