Open GoogleCodeExporter opened 9 years ago
I just saw this issue; it's an interesting one, as those temporary files are
created by the GWT compiler (which GPE invokes).
I think the problem is that on Windows, this directory is never cleaned up. I'm
not sure why it causes Eclipse's startup to slow down though..
We can investigate this one further.
Original comment by rdayal@google.com
on 4 Oct 2012 at 7:01
That would be great!
Original comment by tucker...@gmail.com
on 4 Oct 2012 at 7:14
Yes, I think only unices clean the TMP.
As I wrote, I am not sure that those files are a cause of slowness for eclipse.
But I believe the other reporters.
Anyway, this is at least wasted disk space therefore, my little osgi plugin is
useful (and now it does not need to be started by hand anymore)
Feel free to add it to your next distribution.
If so tell me so that I celebrate the googlization of my simple code ;)
Original comment by wad...@gmail.com
on 6 Oct 2012 at 7:48
Issue 118 has been merged into this issue.
Original comment by danku...@google.com
on 17 Apr 2013 at 7:23
Rajeev: the GWT Compiler and DevMode clean after themselves IFF the JVM is
terminated cleanly. I think this is the main issue: GPE seems to abruptly
terminate the DevMode rather than sending a message asking it to terminate.
Original comment by t.broyer
on 17 Apr 2013 at 9:24
But how about the Terminate and relaunch scenario? I'm (and possibly a lot of
others) using it when the VM has not enough memory or if dev mode starts to be
slow (after an hour of debugging/reloading) since its faster and more
convinient when I'm in the debug perspective. I think Dev mode should keep
track of all temp files it created and clean them in the next run at least. Or
there could possibly be some debug hook in GPE that handles the case when the
Dev mode terminates? Also it happend to me on numerous occasions that the JVM
crashed resulting in registry amd stack dumps (typicaly some sort of heap or
stack error probably due to lack of memory).
Original comment by rame...@gmail.com
on 18 Apr 2013 at 6:43
Let me repost my comment from Issue 118:
I really really feel that shutdown hook is not enough. There are number of ways
that hosted application could be terminated, for number of reasons. One reason
could be fatal error in JVM.
Startup hook should also check for previous startups and clean them up - in
separate thread, of course.
Original comment by rocky...@gmail.com
on 18 Apr 2013 at 10:12
About slow Eclipse startup:
My team members have experienced that behavior when %TEMP% gets cluttered with
previous GWT startup artifacts. My guess is good as anybody else's but I guess
that Eclipse for some reason enumerates the contents of %TEMP% for the reasons
unknown.
Original comment by rocky...@gmail.com
on 18 Apr 2013 at 10:16
Original issue reported on code.google.com by
wad...@gmail.com
on 3 Jul 2012 at 11:56