Re: [Ranger-users] the "preview" branch

[Roman Z.]

On Thu, Oct 07, 2010 at 01:13:24PM -0400, David Barnett wrote:
> I had some questions about the "preview" branch, I asked Roman on github and
> am continuing the discussion here.
> 
> >> So, is it a bug that I got a text dump of images and archives when
> preview_script wasn't configured?
> > Well, without a preview_script, it just shows the raw data which is
> useless to most people and is hidden in ranger-master.
> Right, what I was asking was if the preview settings in
> ranger/fsobject/file.py should maybe be orthogonal to the preview_script
> setting in options.py, so you didn't have to edit both to switch between no
> preview and a script-generated preview. I'm not sure how much in flux those
> settings are, though.

Hm yes, its broken.  Don't know the details now :/ but I think I just
deleted the algorithm and made every file previewable, so it doesn't
disturb the working of scope.sh.  Gonna clean that up of course.
 
> >> Also, I couldn't find anything that was noticeably slow on my machine.
> Was it a certain filetype in particular for you, or did it just slow
> everything down?
> > What's slow is img2txt. It takes over a second to generate previews of
> medium-sized images. Is it just because of my slow computer? (1,6ghz intel
> atom) =[
> It's all pretty much instantaneous for me, but admittedly I'm on a pretty
> high-end laptop (ThinkPad T500, Core 2 Duo @ 2.80GHz, 4GB RAM). I went
> looking for huge images to test it with: a 4.4M JPG took 0.2 sec, and a 6.4M
> JPG just returned immediately with a "failed to load" message. I'm using
> caca-utils 0.99.beta16-3 installed through Ubuntu. Weird...

An acceptable amount of time would be more like 0.02 seconds :).
I really value the ability to fluently move through the list of files
when holding the j key.  That's why there is a (pseudo parallel) loader
which loads directories in the background.  I want to utilize it for
image previews too, but I don't know just how, yet.

As I was typing, I remembered the "cp" branch, which copies files in
another cp/mv process rather than with python shutils, while still being
integrated in the loader/taskview/stuff.  Now I just have to use scope.sh
instead of cp, save the output to some cache, and display it once it's
loaded.  Sounds simple now, why didn't I do it 3 months ago?

> And BTW, I'm not trying to pressure you to get the image previews working.
> It's just a curiosity to me.

Tomorrow is the last exam, so I can start rangering again!