[icons] Add social icons

[mbrookes]

• I have followed (at least) the PR section of the contributing guide.

[oliviertassinari]

These changes open the door to a broad topic: should we author our own icons? I think that it's the type of decisions we need to be careful with (type 1, hard to revert).

Cost. Fontawesome has 1648 icons for a given variant. material.io is 60% behind. I would trust Fontawesome for having prioritized the most important ones. That might be all that it's needed. https://materialdesignicons.com/ has more icons, I haven't count, maybe 2k? We document this alternative. Maybe we don't have to author anything. @leMaik has an interesting project leveraging these components.

Value. The situation seems significantly better since the last synonyms iteration, but I think that we should wait it has more data and see if we still can improve the list. For instance, we might realize that the brand icons aren't the most important ones to add: https://github.com/google/material-design-icons/issues?q=is%3Aissue+is%3Aopen+sort%3Areactions-%2B1-desc
I also think that we should comper the % of no results vs the number of page views of the page. This would give an idea of the developers frustration.

A few stats:

• mdi-material-ui 16k/w downloads
• mdi-react 12k/w downloads
• @material-ui/icons 415k/w downloads
• @fortawesome/react-fontawesome 223k/w downloads
• @mdi/react 8k/w downloads

I would propose that we keep a close eye on the no results of the search icon page, before doing anything. cc @mui-org/core-contributors

[oliviertassinari]

A second thought, almost all of the popular icon sets available have brand icons, so even if they are not the most important icons we could add, they are probably still important.

There is one important aspect that I have overlooked, brand icons name and shapes are supposed to be invariant, it should be safe to add them, we should be able to keep them for a decade.

[oliviertassinari]

Disclamer, adding the brand icons wasn't the idea of Matt, this thread is the results of my self doubts and reflection, blame me, and only me.

[mbrookes]

I think that we should wait it has more data and see if we still can improve the list

The data came from the Material Design Icons top 5000 search result that @Templarian kindly shared on Twitter:

https://t.co/Glp3hiFv8v This is the top 5000 search results for the last year. That might actually be helpful. We track search -> icon click. Early on we used this to figure out icons to make, but not as helpful after 4k.

— Austin Andrews (@Templarian) August 21, 2019

Here's the breakdown:

Query	Rank	Searches	Percent
facebook	32	24,365	0.50%
twitter	79	12,773	0.26%
instagram	86	11,823	0.24%
youtube	137	7,534	0.15%
linkedin	141	7,170	0.15%
apple	213	4,363	0.09%
whatsapp	191	4,972	0.10%
telegram	385	2,124	0.04%
pinterest	574	1,220	0.02%
reddit	831	740	0.02%
snapchat	1566	285	0.01%
google	65	14,919	0.30%
android	261	3,503	0.07%
iphone	444	1,774	0.04%

I've aggregated stubs that are strongly associated with the full name, but not those that could be associated with other icons, or are full words in their own right (tele, pin, link etc.).

For comparison are google, android and iphone that there are already icons for in the Material Icons set.

One thing in favour of adding social icons is that for other UI elements, there is usually a reasonable substitute to be found. But brands are brands, so for a user hoping to use a single source for icons, and needing social icons, these will help. Of course they could always use the paths taken from MDI along with the SvgIcon component.

I don't think that this should automatically open the door to adding a multitude of additional icons.

[oliviertassinari]

@mbrookes Thanks for the stats, very valuable! Ok, I think that we should add them for convenience and because it's pretty safe :).

I don't think that this should automatically open the door to adding a multitude of additional icons.

Agree

[Templarian]

You may find this useful also. We came to a decision roughly a year ago to do a large brand deprecation as part of v5. The roadmap doc below outlines that process.

https://dev.materialdesignicons.com/roadmap/brand-icons

[oliviertassinari]

@mbrookes yarn src:icons starts by removing all the files in the folder the new brand icons have been added. I'm updating a bit the pull request to account for this constraint. Also, the barrel index needs to be updated.

[mui-pr-bot]

No bundle size changes comparing 833738d...bc7dbfb

Generated by 🚫 dangerJS against bc7dbfb

[mbrookes]

yarn src:icons starts by removing all the files in the folder

I had assumed that this was fixed in #17259, since that retains the icons removed by Google in the latest release. If not, then we need to accommodate those too.