[rfc] [icedtea-web] singletons logic, logs and test cleanup/fixes

[Jiri Vanek]

On 12/19/2013 04:42 PM, Andrew Azores wrote:
> On 12/17/2013 10:29 AM, Jiri Vanek wrote:
>> hi!
>>
>> Motivation - after "console aware of plugin" patch, some tests started *sometimes* die as "can not intialize x11" or "null" (not NPE, but just "null" it was that test was null - so class was not even statically instantiated as I found later).
>>
>> Imho the "console aware of plugin" was not the cause, but make this error more visible. Unfortunately i have not found the exact cause, but made few changes which make logic more direct and from those (five) I think two were the cause of "null" and "x11" errors.
>>
>> The rewritten console pathc on review have to be adapted to this.
>>
>> fix 1: initialisation of console gui done in time when gui is shown, not when console is first time "used" - this is crucial for "x11" error.
>>
>> fix2: the NoStdOutErrTest.java methods do not throw any Exception now. Instead of this they log it and pretend to be ok. This was crucial for "null" error. Sometimes race condition or "x11" error (or null was causing x111 ???) throw exception here, which confused junit bloody much. a) it was consumed b)test was not working
>
> Good!
>
>>
>>
>> fix3: removed internal MessageWithLevel class. It lost it sense with addition of headers.
>
> Nice :)
>
>>
>> fix4: debug messages from plugin are not going directly to console output, but are buffered in already exisiting messageQue. I'm not sure how good idea it is. It can lower the synchronisation issues.
>
> Hmm... I think this is okay. It makes sense to use the existing queue.
>
>>
>> fix5: all static factory methods made synchronised. Imho it is the only correct, but.... :) Also I made the synchronisation via "synchronised" but I'm not sure how correct it is... Maybe via static final Object lock = new Object(); is better?
>>
>
> I think using the synchronized keyword is good enough here. If somehow we find later on that it's causing problems then it can be fixed, but I don't think it should be an issue.
>
> Ok for head IMO.
>
>
> Thanks,
>

tahnx, pushed.
I have improved all synchronisation by holder pattern as we spoke yesterday. I will forward this also to 1.4


J.