Re: [Ranger-users] review of "efficient_w3mimgpreview" branch

[Joshua Landau]

On 30 April 2013 02:23, Roman Z. <address@hidden> wrote:

I made a change to the image previewing code which *should* make it
considerably more efficient.  However, before I merge it into master,
I'd like as many people as possible to test it because image previews
produce drastically different results on different computers.

Nice! This reduced CPU load about 40% and does visibly increase render speed.

But it's not quite right, for two main reasons.

The first, and most important, is that rendering a previously-rendered image can be incomplete, depending on environment. I have no test cases to share but some things reproduce it consistently and others don't. Try putting an image at the bottom of a directory about 10 big with a few folders in, each of with a few files, and pressing PgUp (to directory at top) and PgDn (to image). The directory preview should interfere with the image.

The second is that it slows scrolling more than the other. It may just not be a backgrounded process, or something. Because it's so much faster, though, it still comes out ahead for typical images. Brute-force scrolling down for 10 counted seconds got me to the 108th before and the 172nd now. However, viewing a large image seems to cause larger lags than before. There are some very large images I have that now cause ~3/2-second lags whereas they were ~1/3-second beforehand.

What exactly I did was spawn one single w3mimgpreview instance (once
needed) and keep it alive instead of spawning a new one each time an
image is displayed.

 Have you considered caching and pre-rendering? Those would be more significant, á mon avis.