[Laszlo-dev] Anyone else seeing these cache errors?
P T Withington
ptw at laszlosystems.com
Tue Apr 8 20:16:07 PDT 2008
Yikes. Good point. At the very least, maybe we'd bette have them use
separate directories. Which should be easy now... :)
On 2008-04-08, at 18:15 EDT, Donald Anderson wrote:
> On Apr 8, 2008, at 5:50 PM, P T Withington wrote:
>> On 2008-04-08, at 08:53 EDT, Donald Anderson wrote:
>>> On Apr 7, 2008, at 6:50 PM, P T Withington wrote:
>>>> lzunit-basedatacombobox.lzx:1:59: file not found: /Users/ptw/
>>>> 1207603704614.swf (No such file or directory). IOException in
>>>> I keep getting them. Clearing the cache makes them go away, but
>>>> I'm wondering if we broke something recently trying to implement
>>>> the changeable cache dirs?
>>> Possible to be sure, but I've looked at the changes again, and
>>> don't see
>>> what has changed when you don't use these options. I even left
>>> modification in place (in trunk only):
>>> // TODO [hqm 2008-01] I set this to true because when working
>>> // the LFC, I frequently get screwed by using cached script.
>>> boolean flushScriptCache = true;
>>> That should flush the cache every time you run lzc.
>> Right, but I am seeing the error when loading through the server
>> into a browser, not with lzc. Maybe not related, but a surprising
> Hmm. As you know, the recent code change is in compiler/Main.java,
> should only affect running lzc.
> Although mixing lzc runs with browser runs could make it relevant.
> How about this? The script cache for the browser points to the same
> place as the script cache
> when used by lzc. (they are both under the lps directory). BUT --
> they are running in two different
> jvms, and one of these (lzc) is hardwired to clear the cache each
> time. Ooops.
> browser: compile foo.lzc puts foo.swf into cache
> lzc: compile bar.lzc CLEAR cache, and then compile and
> put bar.swf into cache.
> browser (still running): compiler foo.lzc, look for it in the
> cache, it's gone!
> I think it's time for that 'flush the cache' hack to be removed.
> The appearance of the bug is probably coincident with the merge of
> devildog, now
> we all are more often running both lzc and browser out of the same
> - Don
> Don Anderson
> Java/C/C++, Berkeley DB, systems consultant
> voice: 617-547-7881
> email: dda at ddanderson.com
> www: http://www.ddanderson.com
More information about the Laszlo-dev