RFR 8243572: Multiple tests fail with assert(cld->klasses() != 0LL) failed: unexpected NULL for cld->klasses()

Markus Gronlund markus.gronlund at oracle.com
Wed Apr 29 17:24:59 UTC 2020


Hi Harold,

Code changes looks good for fixing the problem, thank you for taking care of this.

Markus

-----Original Message-----
From: Harold Seigel 
Sent: den 29 april 2020 19:13
To: hotspot-runtime-dev at openjdk.java.net
Subject: Re: RFR 8243572: Multiple tests fail with assert(cld->klasses() != 0LL) failed: unexpected NULL for cld->klasses()

Hi,

The purpose of this change is fix the frequent CI tier 7 failures and it is being derailed by the 'hidden' terminology discussion. I'd like to push the current fix (which does not do any renaming of 'hidden') and open a new JFR RFE to deal with 'hidden' terminology.

Does that sound okay?

Thanks, Harold

On 4/28/2020 11:59 PM, David Holmes wrote:
> On 29/04/2020 9:37 am, Mandy Chung wrote:
>> OK.  I can go with "weak hidden" in JFR description as it's informal.
>
> People will go back to the Hidden Classes JEP to try and discern what 
> a "weak hidden" class is. It doesn't make sense to me reject use of 
> "weak hidden" in the core functionality (ie the JEP) and yet allow 
> "informal" use of "weak hidden" elsewhere - it will just raise more 
> questions than it answers IMO.
>
> David
>
>> Mandy
>>
>> On 4/28/20 2:59 PM, Markus Gronlund wrote:
>>>  “Hidden” genera (default) “Strong Hidden” species?
>>>
>>
>> Weak is the default.
>>
>>> If need to make explicit, “Weak Hidden” vs “Strong Hidden”
>>>
>>> “Weak” as a term induces the, historically intuitive, idea of 
>>> not-strong. “Regular” and “Normal” are too general for this concept, 
>>> especially as Hidden Classes are introduced to be an alternative to
>>>  (historically) “Regular” and “Normal” classes.
>>>
>>> 2 cents
>>>
>>> Markus
>>>
>>>
>>>
>>>
>>>
>>> Sent from my iPhone
>>>
>>>> On 28 Apr 2020, at 23:12, Mandy Chung <mandy.chung at oracle.com> wrote:
>>>>
>>>> 
>>>>
>>>> On 4/28/20 1:13 PM, John Rose wrote:
>>>>> On Apr 28, 2020, at 1:10 PM, Mandy Chung <mandy.chung at oracle.com 
>>>>> <mailto:mandy.chung at oracle.com>> wrote:
>>>>>>
>>>>>> "non-strong" is the best term I can come up with.
>>>>>
>>>>> If strong is the non-default choice, then any of “regular”, 
>>>>> “normal”, or “weak” would be OK in my book.  I know “weak” is no 
>>>>> longer a technical term, but as an informal opposite to “strong” 
>>>>> it would work, now.
>>>>>
>>>>
>>>> "regular" or "normal" is a good one.   I didn't suggest that 
>>>> because we use "normal class" to refer to non-hidden class. For 
>>>> this specific discussion about JFR user-visible description, 
>>>> "regular/normal hidden classes" is probably better.
>>>>
>>>>> (This is a big sign of progress:  There’s little remaining to 
>>>>> discuss except bike shed colors!)
>>>>>
>>>> Indeed.
>>>>
>>>> Mandy
>>


More information about the hotspot-runtime-dev mailing list