Make autosave not block on other threads

[naoliv]

(I really don't know JOSM internals, so I will say what I imagine here)

Is it possible to have autosave running in another thread?
For example, after hitting #13142 I can see that JOSM is still reading my OSM messages, reloading modified styles, etc:

INFO: GET https://api.openstreetmap.org/api/0.6/user/details (obter o número de mensagens não lidas) -> 200 (1,35 kB)
INFO: Map style Indoor has been modified. Reloading style...
INFO: GET https://api.openstreetmap.org/api/0.6/user/details (obter o número de mensagens não lidas) -> 200 (1,35 kB)

But it isn't autosaving what I was doing.

If possible, having the autosave in another thread could possibly save whatever I have edited from the last autosave run up to the crash/deadlock/something else (just needing to wait for the next autosave run).

JOSM:

URL:http://josm.openstreetmap.de/svn/trunk
Repository:UUID: 0c6e7542-c601-0410-84e7-c038aed88b3b
Last:Changed Date: 2016-07-10 14:56:47 +0200 (Sun, 10 Jul 2016)
Build-Date:2016-07-11 01:33:05
Revision:10525
Relative:URL: ^/trunk

Identification: JOSM/1.5 (10525 pt_BR) Linux Debian GNU/Linux testing (stretch)
Memory Usage: 247 MB / 10206 MB (157 MB allocated, but free)
Java version: 1.8.0_91-8u91-b14-3-b14, Oracle Corporation, OpenJDK 64-Bit Server VM
VM arguments: [-Dawt.useSystemAAFontSettings=on]

Plugins:
- Create_grid_of_ways (32309)
- ImportImagePlugin (32326)
- OpeningHoursEditor (32309)
- PicLayer (32329)
- RoadSigns (32309)
- SimplifyArea (32309)
- apache-commons (32309)
- buildings_tools (32309)
- conflation (0.2.0)
- contourmerge (1014)
- download_along (32309)
- editgpx (32309)
- ejml (32309)
- geojson (41)
- geotools (32309)
- importvec (32309)
- indoorhelper (32326)
- jts (32311)
- log4j (32309)
- measurement (32324)
- merge-overlap (32309)
- opendata (32351)
- pbf (32309)
- pdfimport (32309)
- poly (32309)
- reverter (32309)
- scripting (30723)
- todo (29154)
- turnrestrictions (32386)
- undelete (32309)
- utilsplugin2 (32333)

[michael2402]

The autosave task is already running on a separate thread ("Timer-0").

The problem is that it interfered with the main thread and deadlocked. The particular problem you encountered was caused by the autosave task running in an other thread.

[michael2402]

I changed the description to better reflect the real problem: The autosave thread depending on other threads.

We should especially avoid the UI thread blocking the autosave thread.