use main lodash instead of separate packages per function

[gazs]

One of our low-hanging fruits is the duplication of lodash imports. Our codebase currently has 4 ways and I'm trying to get it down to 2 (see https://github.com/conde-nast-international/voguede-autopilot/pull/823 ). This is an attempt to fix the upstream instead of trying to rewrite everything on our end.

[pgoldrbx]

I think this lgtm

@rufus2021 who will remember some things I think

cc @CondeNast/autopilot

[pgoldrbx]

probably fine, though we could just specify ^4 to allow maximum deduping downstream?

[pgoldrbx]

Ok, i've done a search, and the lowest common version is 4.14.1

https://github.com/search?l=JSON&p=2&q=org%3ACondeNast+jsonmltoreact&type=Code

• autopilot-ad
• autopilot-teen-vogue
• autopilot-self

cc @dcolucci

[giles-v]

So one of the nice things about this lib is the small size. You're probably juicing it by 15kB or something with this change. Is the assumption that brand apps will always be using the full lodash?

[pgoldrbx]

@giles-v right - the trouble is we either need to always use only the exact lodash packages needed, or we need to always use the full, everywhere, so that we can dedupe in the bundle. If we have a combination, the result will always be larger.

I think there's another thread somewhere about this

[brokentone]

This is excellent, thank you @gazs -- thought I can't see the referenced repo.

@giles-v -- yes I think we have to make this assumption at this point -- and theoretically, it still could be treeshaken downstream. With different packages, it cannot be.

[gazs]

@giles-v this would only increase the build size for brand apps that didn't use lodash at all or were only using the lodash.foo named modules. If they were using lodash/foo, this should be a size reduction for them, if they are importing the monolithical lodash, this should be a size-neutral change (getting rid of lodash.foo and the other lodash modules it brings in )

[jsocol]

I'd really like to see a before/after bundle size comparison from a couple of existing brands before making this switch.

Marking this as "requesting changes" because we can get this data, it's not something we have to guess about. I haven't looked into the code and I'll dismiss this when we know the numbers.

[rufus2021]

could this could result in two versions of the same method for some apps?

[rharriso]

@rufus2021 it could, but even if it isn't resolved by npm, you can force a single version of lodash with webpack

[mize]

Aside from James' timing question, would peerDependencies be useful here? I know we've used the trick in other libs of adding a package to both peerDependencies (when used by a dependent lib/app) and devDependencies (when used for this lib's local dev).

[pgoldrbx]

Yeah, I've considered peerDependencies as a solution as well

[brokentone]

This is a great improvement which I think we should take at this point.

[Juan-M]

FWIW, 👍 too.

[pgoldrbx]

Still blocked on the bundle size change?

[pgoldrbx]

This got lost a long time ago, but really it should have landed. Let's make it land!

[rharriso]

👍

Looks like I'm still part of some team. Which is good, I'm planning on using this in a project coming up here.

[diffcunha]

Good to know this is still being used! 👍

[ajkim]

💪

…

On Wed, Feb 26, 2020 at 2:59 PM Diogo Cunha ***@***.***> wrote: Good to know this is still being used! 👍 — You are receiving this because you are on a team that was mentioned. Reply to this email directly, view it on GitHub <#7?email_source=notifications&email_token=ABNRY5344Q3BHBQIH2GRXN3RE3CYRA5CNFSM4D5RVOQKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOENBVF6I#issuecomment-591614713>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABNRY53ZSMQZIPS2OOW33J3RE3CYRANCNFSM4D5RVOQA> .

[mf2492]

💯

[JulieBusch]

I see I'm not the only one who was mildly surprised to see a Conde email from Github. I applaud your efforts, Phil!

…

On Wed, Feb 26, 2020 at 3:06 PM Michelle Austria Fernandez < ***@***.***> wrote: 💯 — You are receiving this because you are on a team that was mentioned. Reply to this email directly, view it on GitHub <#7?email_source=notifications&email_token=AFL55PL2QOST7GAAZERTQL3RE3DVLA5CNFSM4D5RVOQKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOENBV65Q#issuecomment-591617910>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFL55PJ3GOLJOSVAFG27W6DRE3DVLANCNFSM4D5RVOQA> .

[bencmilton]

[gkilian]

Nothing like a Conde reunion inside a PR!!!

[bmullan91]

Good lad Phil 💪

[brokentone]

I thought more lodash was always better. E.g. at Condé I made a brute force object property extractor leveraging lodash! https://gist.github.com/brokentone/9639f8e0a073aec66ae3b3c67cdf5d9a

[taveras]

👏 👏

[Juan-M]

Wait for me!!!

[poscar]

👋

[Georgette]

GeGe was here.

[Georgette]

[fesebuv]

✅ Well done sir! @pgoldrbx

[whyisjake]

Well done, everyone.