This should be considered a really low priority. We ended up just re-downloading the latest file when the hash changes, rather than using differencing.

With the changes in D9130, this is less relevant than it was originally. In particular, we don't need to be doing any special storage for aggregated lint or coverage data, as this data can be sent directly to Fact and stored generically.

Plan of attack for this task:

The solution for us was to put the command in a batch file and then execute the batch file. I'm not at work right now so I can't confirm whether that fix actually resolves the problem we had.

I'd expect that we'd be using the garbage collecror mechanism to clean up records that are older than X commits or X days. No-one needs this information forever, but you do want to be able to calculate recent trends (is our technical debt increasing or decreasing?)

Nah it's not that, it's more that even administrators can't view the history of a removed comment, so if someone wants to find out what was written in a deleted comment, I've got to start doing SQL queries to find it out.

I think it'd be better to have policy controls on removing comments as well, especially since it removes the ability to view the history of the comment.

Oh derp.

This is the standard package manager on openSUSE, from the official openSUSE repositories.

Installing that package resolves the issue and now https://rpcdn.com.au/file/data/ut6jzdnybslrr66qcrly/PHID-FILE-zi3wo2v4xb4dfhlflwwa/OPTIONS.asset?download=1 works.

Well, this is interesting:

This can be triggered by going to:

For me, in the context of using Phabricator it makes sense for it to be required. I don't really mind have it as a required field when signing up for Phabricator.

Yeah, the JSON blobs in Phabricator use "userdata" so I'd expect Arcanist to change to that.

We can. This means that duration and extra will also get stored in Phabricator (which would be useful for displaying the duration of each test in the UI in the future)

I'm using the __pageview patch and routing another domain through Nginx to
the login / register pages. You can see it at
http://discoverunearth.com/auth/register.

C# would be good. The C# compiler can be configured to automatically build an XML containing all of the documentation, class and method specifications, which Diviner could then read in.

@aik099 Yes, Harbormaster is implemented to some degree, but it's not production ready (I use it in production, but I'm crazy / implemented part of it / both).

Top-level wikis would not be particularly useful for our case (where we have public documentation with internal notes).

But it would be a problem for those of us that would like to have internal subpages of wikis; we don't want to publically announce that there are non-public notes on things because then that makes people ask "what is in the internal notes?"

(Note that the configuration requirement for Phragment is not going to go away; @epriestley and I discussed it at length during the inital construction and it's the only secure way of serving up the dynamic ZIP files)

Closed by commit rP51c4f697f994.

It'd probably be nice just so that clicking "Phabricator" in the top-left doesn't force the user to login (with nowhere else to go).

@epriestley Just following up on this since I had to upgrade today and resolve conflicts (I currently have these patches applied to my install as I need them to run websites). Given that we haven't heard back from Facebook regarding this issue, is there anything blocking it's acceptance in upstream?

Reassigning as I won't get a chance to look at this in the near future.