- Internal Changes to how threads are started and killed
- Added a timeout + retry to calls to IRIDA, so dropped conections will be retried.
- Updated the
requests
library for security patch - Changed the api layer to use a singleton so multiple instances of the api are not spun off.
- Small bug fixes for GUI
- Fixed UI hanging when recursive directories are searched for runs.
- The UI is completely rewritten to better show what's happened with samples as they're uploaded, and to facilitate quality control implementation later.
- Add an experimental auto-upload feature to automatically upload new runs as the sequencer finishes a run (monitoring run directories for
CompletedJobInfo.xml
). - Use the sample name and ID columns correctly in
SampleSheet.csv
: the sample name column is used for naming.fastq.gz
files, the sample ID column is used for sending data to IRIDA. - Discard the cache of projects whenever sending data to the server.
- Add an about dialog to show the version number in the UI.
- Changed the way that update notifications are shown: don't have an open in browser button, just create a link.
- Changed the installer to uninstall the previous version before installing the new version (user settings are preserved).
- Use the
pubsub
built into wxPython rather than pulling in an external version.
- Changed the
Makefile
so that the uploader can be hacked on in other Linux distros, specifically Arch (thanks to @eric.enns) - Fixed a UI issue where upload labels were overlapping in some cases (thanks to @eric.enns)
- Fixed an issue where the uploader found files that had
SampleSheet.csv
anywhere in the filename (i.e.,old_SampleSheet.csv
), now it matches exactlySampleSheet.csv
. - Changed the label
Upload speed
toAverage upload speed
to more accurately reflect the value we're reporting. - Added support for resuming uploads on failure. If the uploader fails when uploading a run, it will now skip any files that were already uploaded.
- Added better error reporting when the uploader can't find the files for the sample by including the directory to help find issues.
- Fix a performance regression when scanning directories for fastq files (thanks to @eric.enns)
- Use pytest for testing instead of the custom testing framework build arount
unittest
. - Add a feature to check the github repository for new releases.
- Add a menu File > Exit menu for familiarity.
- Use a wx widget for directory selection instead of building our own.
- Sample sheet parsing happens with a state machine instead of guessing sections based on number of columns.
- Extensive refactoring to move application functionality out of the GUI layer.
- Added caching for some responses from the server so that we don't have to do a round-trip for every sample.
- Added default directory setting to settings panel.
- Limiting depth of directory scanning to 2 levels.
- Fixed a bug where empty fields in the header metadata section were being incorrectly parsed as read lengths.
- Simplify searching for
SampleSheet.csv
usingos.walk
. Fixes a bug where Windows 7 junction points were being followed, and the OS was reporting permission denied errors on the junction points. This came from auto-scanning the default directory, and the default directory defaulting to the user's home directory. - Add a license!
- Fix an issue where samples that have a common prefix would upload the wrong files for each sample (i.e.,
Sample1
,Sample11
,Sample111
would all upload files forSample111
).
- Fixed some bugs with trailing Windows newlines and trailing commas that get added to header information by Excel.
- Fixed a bug with the settings dialog where when focus was lost on a field the display would get messed up.
- Added a feature to automatically scan the default directory when the app starts up.
- Upgraded to wxWidgets 3 series, and fixed a bug where selecting directories included the drive label.
- Changed the installer to use
pynsist
instead of by-hand construction in Windows with NSIS.