Or, to just download the current version of D25705 to eventually add some local extra changes, just:

If you would like to download the current version of D25705 to maybe land that, just:

In D25705#19508, @valerio.bozzolan wrote:

OK try this to restore your desired situation (so, two diffs)

git stash save

git checkout -b just_phriction
git reset --hard origin/master
arc patch --nobranch --diff 2173
arc diff --update D25705

git checkout -b just_conpherence
git reset --hard origin/master
arc patch --nobranch --diff 2178
arc diff --create

Then you have this benefit that you can repeat as many times you want:

git checkout just_phriction
LOL DO CHANGES
arc diff

git checkout just_conpherence
LOL DO CHANGES
arc diff

Or, get fresh master again:

git checkout master

Ideally, the default assumes that people do not touch master but work in branches, each branch a different feature

Remove unused save() method in PhabricatorUserCache

OK. Giving that also the parent::save() will fail, I think yes, feel free to nuke this method to see what happens. lol

In D25645#19609, @valerio.bozzolan wrote:

It would be nice to understand why nothing calls save() 🤔 since I guess nothing calls it, or we would have a crash report.

or am I supposed to $result = id(new ArcanistHardpointFutureList())->newFromFutures($result);? Yeez

P.S. Why does Phriction have the field text_metadata? And why does Conpherence have content_metadata? 🤔

We can avoid phutil_nonempty_string() since PHP pitfalls "Check for non-empty strings" are under control here.

shorter like in D25705

It would be nice to understand why nothing calls save() 🤔 since I guess nothing calls it, or we would have a crash report.

OK. Well, thaanks \o/

I'm really impressed. Tested and it's super-nice, with implicit middle fingers pointed to JavaScript.

Screenshots of {desktop, tablet, phone} width, exposing an arrow only on tablet and phone as we previously made people scroll large distances anyway and to increase discoverability of the feature in these viewports:

Apply phone style also to tablet style because same rendering (only one column)

In T15872#18251, @valerio.bozzolan wrote:

About this:

unset LC_ALL + LC_MESSAGES=C

I think it's not safe since other LC_STUFF can still be read 🤔

About this:

Set cursor: pointer in CSS

This [LANG=C] is supposed to be the default for missing LC_stuff

final defuser \o/ thanks

OK thanks! you persuaded me

To avoid that $xactions[0] we should probably find the desired transaction.

LC_ALL=C is not risky on GNU systems.

Sigh, xdebug... and seems I need to drop the new from the constructor when the class is final.

Thanks my friend, this works perfectly

we can remove an undeclared variable, and add a nonempty-str check

dedicating Conpherence its own diff as mturdus requested

(BTW, in D25698 we started deprecating remarkup.diviner for RemarkupReferenceController)

I think GitHub allows that syntax only in comments 🤔

removing from review queue until the underlying task discussion is settled.

In T15744#18236, @20after4 wrote:

I was unaware of `#0969DA` syntax from github/gitlab.

I was unaware of #0969DA syntax from github/gitlab. I'm not sure if I like that syntax better than {} but I am generally in favor of using the same syntax as other systems in order to converge towards defacto standardization.

OK try this to restore your desired situation (so, two diffs)

OK, I think we have something mathematical and bombproof

In D25705#19485, @valerio.bozzolan wrote:

(Do you need a little help to split the changes in two diffs?)

Updates after comments

https://unix.stackexchange.com/a/87763 is also a useful reference, specifically:

add even more useful links

I must read this:

Papa Beaver's Storytime done. Thanks

Well, to be honest, feel free to generalize the description and commit message to include both changes, since they are the very same thing

(Do you need a little help to split the changes in two diffs?)

Fix file attachment in Conpherence

Per https://caniuse.com/?search=-webkit-transform, transform without a vendor prefix has been supported since Chrome 36 in 2014.
Personally I'd prefer to simply remove old vendor prefixed CSS definitions but that would first require a shared understanding which dates (and/or which browser versions) are "too old" for Phorge code.
See "List of changes, most recent first" on https://www.mediawiki.org/wiki/Compatibility#Browser_support_matrix for a random example.

Thanks again. I've found this interesting, so, added in the code itself:

improvement thanks to dear @l2dy

Update after comments

Updates after comments

Thanks again @20after4 for the efforts.

This documentation is nice but unfortunately the feature introduced in 4d12d014fd9d never worked :( So documenting this as-is may create confusion, at the moment. Next steps:

Updates after comments

Wow 😍 looks promising (untested from my side)

A day later this turns out to be a rather stupid patch. Things should remain as is as javascript:void(0) nowadays may trigger CSP errors.

what will this do?

Also run ./bin/celerity map because it's JS