Adds a level option to gifsicle

[kaspergrubbe]

Oi!

I had an issue with corruption on some gifs with gifsicle. The corruption (not relevant for image_optim) happened after a resize like:

convert funny_image.gif -resize 100x100 funny_image_resized.gif

I've also added the --careful flag from gifsicle so we produce more compatible gifs, but that might have been a mistake, and please notify if you want me to remove it.

Gifsicle actually obeys the flag -O0 so I considered using that instead of removing the flag, but since it isn't documented by Gifsicle I opted for the solution without -O for when level is zero.

[kaspergrubbe]

I've decided if level == 0 it should send the --unoptimize flag.

[toy]

Great! But few notes: as image_optim is for optimizing images, I don't think that level 0 or --unoptimize make sence unless you can think of a good use case. Also --careful flag is for "Some Java and Internet Explorer versions" so can be added as another worker option but should be off by default.

[toy]

Also I would prefer if you can create a feature branch (unless it will require another pull request, I will add notes about contribution to readme) and make travis happy ;)

[kaspergrubbe]

I will make a featurebranch and work on the specs. I was away for the weekend, but I am planning on looking at this during the week, thanks for your feedback, I will get back to you.

[kaspergrubbe]

Hey @toy

I have cleaned the code up, and merged master in.

I don't think I can make it a featurebranch without making a new PR. Do you know anything about these spec failures?

  1) ImageOptim optimize_image optimizes images
     Failure/Error: expect(nrmse(compare_to, optimized)).to eq(0)

       expected: 0
            got: 0.155859

       (compared using ==)
     # ./spec/image_optim_spec.rb:125:in `block (4 levels) in <top (required)>'
     # ./lib/image_optim.rb:205:in `call'
     # ./lib/image_optim.rb:205:in `block in run_method_for'

[toy]

No interpolation happens.
gifsicle shows only warning about getting "#{level}", so I've created an issue (kohler/gifsicle#29) as this should abort execution.

[toy]

@kaspergrubbe Sorry for delay with reply.

That failure means at least that I need to make failure for that spec more descriptive :)
I can't reproduce failure on local - not good, I'll try further.

I've added few notes, also please run script/update_worker_options_in_readme.

[kaspergrubbe]

Hey @toy, no worries for the delay, we are all busy :-)

Specs passing, and your notes are fixed. I've played around with the descriptions and decided than my initial ones were too long, let me know what you think.

[toy]

Nice!
So it seems that spec failure was caused by uninterpolated --optimized argument and different behaviour of older gifsicle (1.64 vs 1.84).

[kaspergrubbe]

So the specs worked, good! :P

Glad to have this merged, thanks!