Google Fonts not working since 2.4.1

I’ll check with the developer who made contribution on Github Clement, thanks for reporting!

Hi @renoovodesign,
would you mind providing me with a link to one of your websites where this issue happens?

Thanks!

Edit:
nevermind Clement, I’m able to reproduce the issue on my end. Thanks!

• This reply was modified 6 years ago by Rocco Aliberti.

so had the wrong developer in mind, thanks for looking into this Rocco! ??

Clement; I put this on my list of things to fix for 2.4.2, I’ll be back here when I have a solution so you can test it, probably later this week.

Did some tests with AO 2.4.0, 2.4.1 and current master and can’t reproduce I’m afraid, I have not seen the subsets being doubled. Moreover the only change for the Google Fonts optimization was the addition of a (previously missing) & as per this changeset on www.ads-software.com trac.

So .. I’m going to need some help on this one so I can at least reproduce @d4z_c0nf and/ or @renoovodesign ? ??

Hi @optimizingmatters
I’ve provided a way to reproduce the issue in
https://github.com/futtta/autoptimize/issues/203

with the Twenty Seventeen theme.
Doesn’t work for you?

OK, so your PR (which will be part of 2.4.2) will fix the missing %C2 (comma), the (remaining) question (which I was too focused on maybe) is why the subsets appear twice in OP’s example?

subset=latin,latin-extlatin,latin-ext in 2.4.1
vs
subset=latin,latin-ext in 2.4.0

Well the “duplication” doesn’t produce any issues as far as I could see and test, the missing comma does, and that’s what I addressed – that was the bug for me ??

Also, I remember I did some rapid tests comparing Google response both when the subset contained duplication and when not and couldn’t spot any difference, hence I’ve decided to not complexify the code, and let Google “uniquify” the subset.

I might be wrong though, and in this case we can just ‘explode’ the subset to an array and go with array_unique and then ‘implode’ it again… But again I couldn’t find the duplication as source of any issue.

My 2 cents. ??

p.s.
Edit: but of course you know better than me how to fix it.
I’ve just realized that I didn’t answer to your question which was “why does the duplication happen”… The answer, to me, is just that the duplication isn’t something Autoptomize produces, the duplication happens because the fonts are enqueued by several different parts, and Autoptimize just doesn’t “strip” duplicates in the subset. So did Autoptimize strip duplicates before, if yes when and what changed ever since? I’m sorry I don’t know :/

Can look furthermore into it as soon as I can.

• This reply was modified 6 years ago by Rocco Aliberti.

Some sound arguments Rocco, thanks! So unless Clement confirms the duplication is due to AO and that it is creating problems even if the encoded comma is added (as per your PR) I agree we can just leave this be knowing your PR fixes the major issue here.

thanks again!
frank

From a user point of view the duplication of subsets doesn’t really make sense since the options read “Combine and …”
However if technically there is no difference and the fonts load properly then it sounds fine to me.

thanks for the feedback @renoovodesign.

could you:
* download & test the github version of what will become AO 2.4.2 to confirm if the problem (HTTP 400) is fixed?
* share the URL of a site that had this problem so I could try reproduce the duplication?

frank

@optimizingmatters

in this example:
https://github.com/futtta/autoptimize/issues/203#issue-370127006

if you replace ‘vietnamese’ with ‘latin’
you should reproduce the subset “duplication”,
since by default twentyseventeen already requires the latin subsest for its Libre Franklin.

Rocco

@optimizingmatters, I have tested the github version and it has fixed the 400 error. Many thanks.

OK, thanks for the feedback Clement ??