diff options
author | 2015-05-27 17:06:27 -0400 | |
---|---|---|
committer | 2015-05-27 17:06:27 -0400 | |
commit | e94a3e5a9faf7f6795eaf8d31576884e3b34f143 (patch) | |
tree | 31a8a5acc697e89e391bbb6e0f6d7fc56876c3d6 /doc/devblog | |
parent | d7a275122be10db6cf2f2877a5f15659177684ee (diff) |
devblog
Diffstat (limited to 'doc/devblog')
-rw-r--r-- | doc/devblog/day_288__microrelease_prep.mdwn | 22 |
1 files changed, 22 insertions, 0 deletions
diff --git a/doc/devblog/day_288__microrelease_prep.mdwn b/doc/devblog/day_288__microrelease_prep.mdwn new file mode 100644 index 000000000..fc1a5f3df --- /dev/null +++ b/doc/devblog/day_288__microrelease_prep.mdwn @@ -0,0 +1,22 @@ +After a less active than usual week (dentist), I made a release last Friday. +Unfortunately, it turns out that the Linux standalone +builds in that release don't include the webapp. So, another release is +planned tomorrow. + +Yesterday and part of today I dug into +the [[bugs/windows_ssh_webapp_password_entry_broken]] +reversion. Eventually cracked the problem; it seems that +different versions of ssh for Windows do different things in a `isatty` +check, and there's a flag that can be passed when starting ssh to make it +not see a controlling tty. However, this neeeds changes to the +`process` library, which db48x and I have now coded up. So a fix for this bug +is waiting on a new release of that library. Oh well. + +Rest of today was catching up on recent traffic, and improving the behavior +of `git annex fsck` when there's a disk IO error while checksumming a file. +Now it'll detect a hardware fault exception, and take that to mean the file +is bad, and move it to the bad files directory, instead of just crashing. + +I need better tooling to create disk IO errors on demand. +Yanking disks out works, but is a blunt instrument. Anyone know of +good tools for that? |