Commit Graph

8221 Commits

Author SHA1 Message Date
Jean-Paul Calderone
b76d23e45a news fragment 2019-03-18 16:36:48 -04:00
Jean-Paul Calderone
262485bd14 add a test for the conflict behavior 2019-03-18 16:35:42 -04:00
Jean-Paul Calderone
9bc00c4f76 add a couple more helpers 2019-03-18 16:35:35 -04:00
Jean-Paul Calderone
a44c5d0f11 respect the timeout 2019-03-18 16:35:31 -04:00
Jean-Paul Calderone
55690bf639 Do it without a new flag
This is safer against other conditions where _deque ends up with items when we
start a scan - for example, if we chunk up deque processing.
2019-03-18 15:38:18 -04:00
Jean-Paul Calderone
6a4d461d38 only break it once because we don't do a full iteration after, now 2019-03-18 15:28:43 -04:00
Jean-Paul Calderone
d3fde4a4c5 and put the iterate into an action 2019-03-18 15:28:35 -04:00
Jean-Paul Calderone
0ac96e9113 put the _process_deque into an action too 2019-03-18 15:28:16 -04:00
Jean-Paul Calderone
20150541a7 put the _perform_scan into an eliot action 2019-03-18 15:27:38 -04:00
Jean-Paul Calderone
e1de96486c Don't re-run _perform_scan in this abstraction violating test
Just do what comes after _perform_scan
2019-03-18 15:11:25 -04:00
Jean-Paul Calderone
d1a62b1c99 Attempt to avoid the startup double-scan and related problems 2019-03-18 14:24:34 -04:00
Jean-Paul Calderone
c01c85522d Use the method
There is a method.  Why not use it?
2019-03-18 14:18:09 -04:00
Jean-Paul Calderone
135462ccd3
Merge pull request #571 from tahoe-lafs/3000.windows-coverage
Report code coverage from Windows CI

Fixes: ticket:3000
2019-03-15 13:25:37 -04:00
Jean-Paul Calderone
6fb1252f0e install coverage here too 2019-03-14 19:25:39 -04:00
Jean-Paul Calderone
0bead3638e invoke via -m 2019-03-14 19:25:35 -04:00
Jean-Paul Calderone
7300125170 "no coverage to combine"!
`coverage combine` fails if you give it no paths and there is no `coverage run
-p` data, it seems.  Try coverage reporting without the combine step.
2019-03-14 16:43:06 -04:00
Jean-Paul Calderone
bc7ae0eabc Perhaps this invocation will result in greater success 2019-03-14 16:18:46 -04:00
Jean-Paul Calderone
f5a97ef1ad correct the indentation to make it valid yaml 2019-03-14 13:26:12 -04:00
Jean-Paul Calderone
c1e6f08813
Merge pull request #558 from tahoe-lafs/1432.watchdog-magic-folder-with-eliot
Add support for macOS to Magic-Folders.

Fixes: ticket:1432
2019-03-14 13:02:09 -04:00
Jean-Paul Calderone
fd49f6f52c news fragment 2019-03-14 11:54:26 -04:00
Jean-Paul Calderone
f4c1038a24 Report coverage information to The Cloud™ 2019-03-14 11:54:00 -04:00
Jean-Paul Calderone
31287efa95 Collect coverage information from the test suite 2019-03-14 11:53:52 -04:00
Jean-Paul Calderone
17c32a874b Fix comment typo 2019-03-14 11:53:45 -04:00
Jean-Paul Calderone
365ff1a02c explain why we need to touch() twice 2019-03-14 11:26:06 -04:00
Jean-Paul Calderone
5141105441 Fix typo 2019-03-14 11:26:06 -04:00
Jean-Paul Calderone
2cb710ac1e Just touch the path once 2019-03-14 11:26:06 -04:00
Jean-Paul Calderone
b14583f0e9 Remove this broken ignore_count thing
The only purpose it serves to make the tests invalid and useless.
2019-03-14 11:26:06 -04:00
Jean-Paul Calderone
dfd0c6e54c fix "downloader" type in message type 2019-03-14 11:26:06 -04:00
Jean-Paul Calderone
3978c45b04 spit out the events as raw as we can 2019-03-14 11:26:06 -04:00
Jean-Paul Calderone
fef79cd4a3 re-use the _add_watch helper 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
fcb08b5a3a Move these Eliot events somewhere reusable 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
9ed019f8de Give the tests a little more time. Windows may need it. 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
64826aee37 Clean up stdout a bit 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
3176b6f18a Create the watchdog Observer sooner
This lets us watch before start which Windows seems to really want.
2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
263755fb26 Watch before we start.
Linux and macOS are happy to watch after starting.  Our Windows support
library can't deal with such a case, though.

Linux is happy with the other order.  I'm about to find out if macOS is.

There are likely further Windows issues to deal with.
2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
2205bf0fdd unused 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
241b6cedfe Remove the skipped tests
They're skipped because we don't need that functionality.  If we need it in
the future, we'll implement it with tests.
2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
984d27ab62 simplify 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
67ccd21481 python syntax 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
35fc563a6f Attempt to work-around watchdog bug 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
e0b7919ad8 macOS is picky 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
4cccf5a316 it sure would be useful to be able to inspect these 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
c5c7f954e9 fix it! 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
515475022a add a test for deleting a file and its containing directory
at about the same time
2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
0822a71163 If it's some _other_ error then propagate it! 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
c2e19cc098 fix a raft of directory handling issues 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
4b0b269b56 Perform the asserts earlier 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
ed755cf590 it didn't start if the path is invalid 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
f2cfad0cb9 There _is_ no traceback. It's an else. 2019-03-14 11:24:35 -04:00
Jean-Paul Calderone
db7af2661a Close a window for a race condition in notification processing 2019-03-14 11:24:35 -04:00