Optimize PNG images

[addyosmani]

We currently auto-generate many of the icons in the repo, but our PNGs could be optimized further. We definitely plan on doing this but want to work it into our asset optimization flow at Google first.

Side: at some point, we may well open-source the tooling process around the icons or add them to this repo but our current focus is on making sure the icons provided are as polished and consumable as possible.

[tomByrer]

our PNGs could be optimized further

Yep, took a minute to compress 6 random PNGs, & got 20-35% reduction in filesize. Even if you had to do it "by hand" (I just drag & dropped a folder into my tool), couldn't you do that before release please?
Perhaps this should have been tagged v0.1.0 instead of v1.0.0?

[addyosmani]

So, I actually did go through and manually optimize the PNGs before release :) In the end we decided to go ahead with the script-generated versions of the icons for V1.

That might sound crazy, but it's because (for now) we want the Google (internal) version of these icons to be the source of truth and minimizing the difference of what is in this repo and there makes it easy to keep the public version up to date.

Soon, we hope to integrate proper PNG and SVG optimization into our pipeline so that everything we sync back to this repo is kept as small as possible (hopefully for 1.0.x). We definitely do care about byte-shaving :)

Perhaps this should have been tagged v0.1.0 instead of v1.0.0?

Nope! We're following npm's more recent semver recommendations, which suggest starting with 1.0.0 and iterating on that baseline rather than opting for the more classic 0.x.y model. We're new to it too, so still working out how well that model works.

[tomByrer]

because (for now) we want the Google (internal) version of these icons to be the source of truth

Ohhh, that explains alot, thanks. Shouldn't this be in the README then? I've learned to look into the issues to see if something is truly ready to go for consumption (or in my case CDN hosting), but most others will see v1.0.0 & immediately start using.

following npm's more recent semver recommendations

Yes, I watched that thread also. I do think people are too afraid to bump up to 1.0.0. But I think v1.0.0 should be 'ready to consume'. Perhaps I'm too much of a perf nerd, but even you agree there needs to be more steps to be truly ready. I would have released as v0.6.0; "We're more than half-way there".

[j9t]

(Seconding @tomByrer’s point. Average savings through ImageOptim are about 25%. Holding off a pull request per the discussion here.)

[tomByrer]

Yep, I used tool a similar to ImageOptim for Windows.
Obviously gulp-imagemin is a non-answer for some, so I hope a more universal tooling is figured out.

[ilanbiala]

+1 optimization would be great at least for the released icons. Maybe have a Travis worker build these and do the release?

[shyndman]

We've ultra optimized everything, and will continue to do so.