ooh, check out D25349 for basically this exact use-case...

Thanks for this patch. Yeah if I understand correctly, this should work:

hsprintf(
  '<tt>%s $</tt> ./bin/auth lock',
  PlatformSymbols::getPlatformServerPath())

And maybe in the future we can have a CSS selector to be applied just to the "phorge/" pathname, in order to avoid mouse selection, and select just the command. But now let's fix just this brand issue.

In D25425#12230, @valerio.bozzolan wrote:

maybe in the future we can have a CSS selector to be applied just to the "phorge/" pathname, in order to avoid mouse selection, and select just the command.

Yeah, I would like to see that. One way to implement it is to produce the prompt character and directory prefix via CSS, as a pseudo-element (::before { content: 'foobar'; }), but it's also possible to wrap the prompt in an addressable element and apply user-select: none; to it. I personally prefer the former approach (it seems more semantically appropriate) but the latter is also valid.

So, we are here if I understand correctly:

1. having the <tt>phorge/ $</tt> ./bin/auth lock (thanks to getPlatformServerPath())
   • waldyrious: 🔴 don't like since the "phorge" is alien confusing text that is selected automatically with a double-click by most browser webs
   • valerio: ✅ Pro: 100% backward compatible. Cons: I agree with waldyrious but I think we can create a CSS class for non-auto-selectable things later in the future. So 1. is good.
   • avivey: ✅ like
2. having <tt>$</tt> ./bin/auth lock
   • waldyrious: ✅
   • valerio 🔴 I don't like since the prefix was added to have arcanist/phorge indication. So, if we use just $ it's better to remove it.

So the compromise:

3. having just ./bin/auth lock without any silly prefix
   • valerio ✅ also this could be expanded in the future
   • waldyrious: ✅ (probably?)
   • avivey: probably 🤷

So we can maybe just propose a version n. 3 and then, later, do a follow-up change introducing:

->addCommandPrefix(getPlatformServerPath())

So that this will take care of the <tt> thing and making that non-selectable etc. Later.

the prefix was added to have arcanist/phorge indication. So, if we use just $ it's better to remove it.

Considering that all the other such messages have a plain $, I'm not sure this prefix addition did fulfill that goal. It currently introduces inconsistency with regards to the other similar messages, which have no path at all (and of course, it is an outdated reference); the inconsistency and the outdated reference is what this patch aims to address. That said, I'd be definitely in favor of a future patch adding the prefix to all messages, using getPlatformServerPath(). I'd be happy to prepare that one myself, even, right after this one.

So we can maybe just propose a version n. 3 and then, later, do a follow-up change

I get where you're coming from, and I do agree that it's the desirable ultimate state, but until the CSS-based prompt marker is implemented, I believe the hardcoded one offers a better experience since it's a common pattern used in many places to indicate a command to be run in a terminal. So I agree with removing the $, but I'd rather do that in the patch that introduces the CSS-based $, so that there's no regression in the information conveyed to the readers of these messages. And again, I'm happy to propose that patch myself, after this one is merged.