Failing test for symlinks

[adgad]

/cc @matthew-andrews

Leaving this here as a failing test for following symlinks in partials dir. It's fixed in my fork of express-handlebars, which I've put in a PR for:
"express-handlebars": "adgad/express-handlebars#follow-symlinks",

https://github.com/ericf/express-handlebars/pull/98/files

[matthew-andrews]

oh wow. really nice work.

[matthew-andrews]

this is exactly the approach I strongly believe in — identify the root cause, fix it there… contribute back to open source, promote better understanding the internals of the tools we use (eg. I didn't know bower link used symlinks — I guess it's obvious now but I never thought about it… just assumed it was “Magic”), regression test, and a scalpel-like fix, really really nice work.

[matthew-andrews]

If you like we can switch to using your branch's express-handlebars in ft-next-express. I guess we will see how long it takes to get the proper fixed merged in upstream?

[adgad]

Yeah let's see where it's at in the new year, if not then we can use the
fork

On 24 December 2014 at 11:01, Matt Andrews [email protected] wrote:

If you like we can switch to using your branch's express-handlebars in
ft-next-express. I guess we will see how long it takes to get the proper
fixed merged in upstream?

—
Reply to this email directly or view it on GitHub
#9 (comment)
.

---

This email was sent by a company owned by Pearson plc, registered office at
80 Strand, London WC2R 0RL. Registered in England and Wales with company
number 53723.

[adgad]

Okay, so the node-glob PR around following symlinks seems to have stalled. Given that the bower link workflow is quite important for us, I propose we operate off a fork of express-handlebars for now, with one of the following hacky solutions:

https://github.com/adgad/express-handlebars/compare/hacky-symlink-fix?expand=1
(I don't know enough about globs to know why that works, but it does)

OR rolling back node-glob to the v3.x branch (which had different symlink behaviour) as suggested here isaacs/node-glob#134.

Thoughts?

[matthew-andrews]

Actually the hacky solution makes sense based on isaacs comment

The solution is not "follow symlinked dirs" as suggested in the OP here. The correct behavior is "stop following symlinked dirs as part of a globstar match, but still resolve their contents against subsequent parts of pattern".

Personally, though, I don't think it's hacky enough #14