[Pharo70 alpha] Store settings raise a DNU primFlush

> On 13 Apr 2018, at 22:26, Alistair Grant <[hidden email]> wrote:
>
> Hi Stef & Sven,
>
> On 13 April 2018 at 20:38, Stephane Ducasse <[hidden email]> wrote:
>> Hi
>>
>> I do not know if this is related to the OldStreamCleaningEffort but
>> when I press store settings in the Settings browser I get a primitive
>> failed.
>>
>> https://pharo.fogbugz.com/f/cases/21699/Store-settings-raise-a-DNU-primFlush
>
>
> This is because the stream created in
> SystemSettingsPersistence>>storeExactStoredSettings: is closed twice,
> first in SettingsStonWriter>>store and then in
> SystemSettingsPersistence>>storeExactStoredSettings:.
>
> Previous versions of file streams have allowed the stream #close to be
> called multiple times, however ZnBufferedReadWriteStream doesn't allow
> this.
>
> I think that since an already open stream is handed to
> SettingsStonWriter, it shouldn't be closing the stream.  That will
> resolve this particular problem.
>
> It can also be argued that ZnBufferedReadWriteStream and co. should
> allow streams to be closed multiple times for backward compatibility.
> Sven?

>
>
>> On 13 Apr 2018, at 22:26, Alistair Grant <[hidden email]> wrote:
>>
>> Hi Stef & Sven,
>>
>> On 13 April 2018 at 20:38, Stephane Ducasse <[hidden email]> wrote:
>>> Hi
>>>
>>> I do not know if this is related to the OldStreamCleaningEffort but
>>> when I press store settings in the Settings browser I get a primitive
>>> failed.
>>>
>>> https://pharo.fogbugz.com/f/cases/21699/Store-settings-raise-a-DNU-primFlush
>>
>>
>> This is because the stream created in
>> SystemSettingsPersistence>>storeExactStoredSettings: is closed twice,
>> first in SettingsStonWriter>>store and then in
>> SystemSettingsPersistence>>storeExactStoredSettings:.
>>
>> Previous versions of file streams have allowed the stream #close to be
>> called multiple times, however ZnBufferedReadWriteStream doesn't allow
>> this.
>>
>> I think that since an already open stream is handed to
>> SettingsStonWriter, it shouldn't be closing the stream.  That will
>> resolve this particular problem.
>>
>> It can also be argued that ZnBufferedReadWriteStream and co. should
>> allow streams to be closed multiple times for backward compatibility.
>> Sven?
>
> Yes, OK, making #close idempotent is good. If we add #closed to ZnBufferedReadStream we can add a
>
>   self closed ifFalse: [ .. ]
>
> guard in #close.
>
>> Cheers,
>> Alistair
>>
>
>

> 2018-04-14 16:24 GMT+02:00 Sean P. DeNigris <[hidden email]>:
> Sven Van Caekenberghe-2 wrote
> > If we add #closed to ZnBufferedReadStream…
>
> Would #isClosed be more consistent with the rest of the system?
>
>
>
> -----
> Cheers,
> Sean
> --
> Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html
>
>