Conversation
7459c15 to
6415fa3
Compare
|
I'll have to figure out what kind of declaration that is and then update the parser. |
|
Interesting. The compiler used by Travis can't find the |
Because you're not specifying with which dmd to build dscanner. It just picks up whatever system-wide dmd installed. Please fix that, e.g. similarly to how we pin stable dmd versions for building dlang.org's ddocs or dmd's bootstrap compiler. |
|
... and now we run into problem that currently our installer is broken for point releases :/ |
96f7ee3 to
2f54a27
Compare
|
Okay I once again updated this PR to include the changes that managed to sneak in over the last month. I hope that this version now passes our CI, s.t. we can finally enable the check again :) |
std/random.d
Outdated
| { | ||
| import std.algorithm : canFind, map; | ||
| import std.algorithm.searching : canFind; | ||
| import std.algorith.iteration : map; |
There was a problem hiding this comment.
Failing on this line. algorithm is misspelled
2f54a27 to
6a0e0b2
Compare
|
I went through another round of catching up & thus rebased to the lastest changes. It would be quite nice if we can get this in rather sooner than later :) |
std/experimental/ndslice/internal.d
Outdated
| } | ||
|
|
||
| auto opBinary(string op)(sizediff_t shift) | ||
| typeof(this) opBinary(string op)(sizediff_t shift) |
There was a problem hiding this comment.
This is technically a false positive, right?
There was a problem hiding this comment.
Yes unfortunately Dscanner can't expand mixins atm, s.t. is the poor man's workaround "for the greater good".
There was a problem hiding this comment.
Oh it seems that Dscanner got even smarter: dlang-community/D-Scanner#379, so I reverted that workaround.
|
Auto-merge toggled on |
|
How come I don't see CircleCI in the CI services list on this PR? |
|
Auto-merge toggled off |
|
Its actually disappeared from all PRs, so probably not due to these changes. |
|
Perhaps it's because of this setting? Link goes to https://circleci.com/docs/fork-pr-builds/. I don't know how CircleCI is configured (heck, I have no clue why we even need it at all considering there's Travis and AppVeyor), so I'm not sure if it's safe to enable or if it's relevant at all. @wilzbach ? |
Oh sorry - I was hoping that it was just a temporary outage and would start to build the PR.
The only reason for using CircleCi is because it's not Travis :P |
So, to clarify... CircleCI has more throughput than Travis, or using both gets us more throughput than using just one, or something else? |
Yes it's solely about having more throughput. We tried to use Travis for Phobos before and the builds were stuck in the queue for days, hence the move to CircleCi. I don't know the reason for the outage with CircleCi yet, but unfortunately it doesn't have anything to do with the "Permissive build setting"). |
6a0e0b2 to
f0c5a9f
Compare
ecfc617 to
0e9f16e
Compare
Seems like you were right. As we don't have any sensitive information stored with CircleCi, I flipped "Permissive building of fork pull requests" to "on" and it seems to be building now :) While I was looking at our config I also realized that a bug in Dscanner has been fixed and we can now re-enable the style checking for |
|
Auto-merge toggled on |

It seems like we/I accidentally disabled the style target on CircleCi (
styleis not defined in the command switch). I take the full blame for this, but at least it's a good proof that the CI actually does catch quite a few things ;-)I split up the fixes depending on their category.
There's only one line left about which DScanner is complaining (hence the "blocked" label):
It warns with:
@Hackerpilot is there a simple fix? Otherwise I will add this file to the Dscanner blacklist.