You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@polygene.apache.org by Niclas Hedhman <ni...@hedhman.org> on 2015/08/26 17:33:37 UTC
Stream/Functional progress report
Gang,
I want to give you an update on how it is going...
In essence, I have completed the first round of massive changes, where most
of the internals are using Stream API instead of the Iterable system from
Rickard.
It compiles and some non-trivial tests pass.
BUT, I must have messed up some of the functions, because withTags()
doesn't work anymore. When the withTags() are to be converted into a
ServiceTags instance and placed into MetaInfo, something goes wrong and
ServiceTags are never added. This breaks a LOT of tests, due to "json"
tagging on ValueSerialization, and I haven't stumbled upon breaking for
other reasons, so fixing this could(!) clear up a massive number of tests
in one go.
I also note that the Build Passes, even though most tests fail. That can't
be right...
I am committing and pushing this anyway, in case someone has time to look
at this detail tonight or tomorrow, as I need to run around with other
things quite a lot, for my move to Shanghai.
So, my guess is that CI will break and start making noise.
--
Niclas Hedhman, Software Developer
http://zest.apache.org - New Energy for Java
Re: Stream/Functional progress report
Posted by Niclas Hedhman <ni...@hedhman.org>.
I have also traced down a couple of more such mistakes.
Out of the ~1300 tests in Core, 80 are now failing... Getting there.
On Thu, Aug 27, 2015 at 9:10 AM, Niclas Hedhman <ni...@hedhman.org> wrote:
> Ah, ServiceTags are not an Interface.... And I misinterpreted the
> flatten() in previous code.
>
> Committing.
>
> On Thu, Aug 27, 2015 at 8:26 AM, Niclas Hedhman <ni...@hedhman.org>
> wrote:
>
>> In fact, line 195 should be adding the top-most interface already.
>>
>> On Thu, Aug 27, 2015 at 8:24 AM, Niclas Hedhman <ni...@hedhman.org>
>> wrote:
>>
>>> Yeah, I kind of figured that much out already. But I still can't spot
>>> the difference from the old code.
>>>
>>> And not comfortable of just adding it.
>>>
>>> On Thu, Aug 27, 2015 at 6:53 AM, Kent Sølvsten <kent.soelvsten@gmail.com
>>> > wrote:
>>>
>>>> Debugger tells that the bug is inside Classes#TYPES_OF.
>>>>
>>>> The old implementation returned both interfaces and parent classes for a
>>>> class, - the new implementation returns only interfaces.
>>>> The effect is that we only store the ServiceTags under metaInfo keyed by
>>>> the Serializable interface, and thus are unable to find it.
>>>>
>>>> Who will be first to fix it?
>>>>
>>>> /Kent
>>>>
>>>> Den 26-08-2015 kl. 17:33 skrev Niclas Hedhman:
>>>> > Gang,
>>>> >
>>>> > I want to give you an update on how it is going...
>>>> >
>>>> > In essence, I have completed the first round of massive changes,
>>>> where most
>>>> > of the internals are using Stream API instead of the Iterable system
>>>> from
>>>> > Rickard.
>>>> > It compiles and some non-trivial tests pass.
>>>> >
>>>> > BUT, I must have messed up some of the functions, because withTags()
>>>> > doesn't work anymore. When the withTags() are to be converted into a
>>>> > ServiceTags instance and placed into MetaInfo, something goes wrong
>>>> and
>>>> > ServiceTags are never added. This breaks a LOT of tests, due to "json"
>>>> > tagging on ValueSerialization, and I haven't stumbled upon breaking
>>>> for
>>>> > other reasons, so fixing this could(!) clear up a massive number of
>>>> tests
>>>> > in one go.
>>>> >
>>>> > I also note that the Build Passes, even though most tests fail. That
>>>> can't
>>>> > be right...
>>>> >
>>>> > I am committing and pushing this anyway, in case someone has time to
>>>> look
>>>> > at this detail tonight or tomorrow, as I need to run around with other
>>>> > things quite a lot, for my move to Shanghai.
>>>> >
>>>> > So, my guess is that CI will break and start making noise.
>>>> >
>>>>
>>>>
>>>
>>>
>>> --
>>> Niclas Hedhman, Software Developer
>>> http://zest.apache.org - New Energy for Java
>>>
>>
>>
>>
>> --
>> Niclas Hedhman, Software Developer
>> http://zest.apache.org - New Energy for Java
>>
>
>
>
> --
> Niclas Hedhman, Software Developer
> http://zest.apache.org - New Energy for Java
>
--
Niclas Hedhman, Software Developer
http://zest.apache.org - New Energy for Java
Re: Stream/Functional progress report
Posted by Niclas Hedhman <ni...@hedhman.org>.
Ah, ServiceTags are not an Interface.... And I misinterpreted the flatten()
in previous code.
Committing.
On Thu, Aug 27, 2015 at 8:26 AM, Niclas Hedhman <ni...@hedhman.org> wrote:
> In fact, line 195 should be adding the top-most interface already.
>
> On Thu, Aug 27, 2015 at 8:24 AM, Niclas Hedhman <ni...@hedhman.org>
> wrote:
>
>> Yeah, I kind of figured that much out already. But I still can't spot the
>> difference from the old code.
>>
>> And not comfortable of just adding it.
>>
>> On Thu, Aug 27, 2015 at 6:53 AM, Kent Sølvsten <ke...@gmail.com>
>> wrote:
>>
>>> Debugger tells that the bug is inside Classes#TYPES_OF.
>>>
>>> The old implementation returned both interfaces and parent classes for a
>>> class, - the new implementation returns only interfaces.
>>> The effect is that we only store the ServiceTags under metaInfo keyed by
>>> the Serializable interface, and thus are unable to find it.
>>>
>>> Who will be first to fix it?
>>>
>>> /Kent
>>>
>>> Den 26-08-2015 kl. 17:33 skrev Niclas Hedhman:
>>> > Gang,
>>> >
>>> > I want to give you an update on how it is going...
>>> >
>>> > In essence, I have completed the first round of massive changes, where
>>> most
>>> > of the internals are using Stream API instead of the Iterable system
>>> from
>>> > Rickard.
>>> > It compiles and some non-trivial tests pass.
>>> >
>>> > BUT, I must have messed up some of the functions, because withTags()
>>> > doesn't work anymore. When the withTags() are to be converted into a
>>> > ServiceTags instance and placed into MetaInfo, something goes wrong and
>>> > ServiceTags are never added. This breaks a LOT of tests, due to "json"
>>> > tagging on ValueSerialization, and I haven't stumbled upon breaking for
>>> > other reasons, so fixing this could(!) clear up a massive number of
>>> tests
>>> > in one go.
>>> >
>>> > I also note that the Build Passes, even though most tests fail. That
>>> can't
>>> > be right...
>>> >
>>> > I am committing and pushing this anyway, in case someone has time to
>>> look
>>> > at this detail tonight or tomorrow, as I need to run around with other
>>> > things quite a lot, for my move to Shanghai.
>>> >
>>> > So, my guess is that CI will break and start making noise.
>>> >
>>>
>>>
>>
>>
>> --
>> Niclas Hedhman, Software Developer
>> http://zest.apache.org - New Energy for Java
>>
>
>
>
> --
> Niclas Hedhman, Software Developer
> http://zest.apache.org - New Energy for Java
>
--
Niclas Hedhman, Software Developer
http://zest.apache.org - New Energy for Java
Re: Stream/Functional progress report
Posted by Niclas Hedhman <ni...@hedhman.org>.
In fact, line 195 should be adding the top-most interface already.
On Thu, Aug 27, 2015 at 8:24 AM, Niclas Hedhman <ni...@hedhman.org> wrote:
> Yeah, I kind of figured that much out already. But I still can't spot the
> difference from the old code.
>
> And not comfortable of just adding it.
>
> On Thu, Aug 27, 2015 at 6:53 AM, Kent Sølvsten <ke...@gmail.com>
> wrote:
>
>> Debugger tells that the bug is inside Classes#TYPES_OF.
>>
>> The old implementation returned both interfaces and parent classes for a
>> class, - the new implementation returns only interfaces.
>> The effect is that we only store the ServiceTags under metaInfo keyed by
>> the Serializable interface, and thus are unable to find it.
>>
>> Who will be first to fix it?
>>
>> /Kent
>>
>> Den 26-08-2015 kl. 17:33 skrev Niclas Hedhman:
>> > Gang,
>> >
>> > I want to give you an update on how it is going...
>> >
>> > In essence, I have completed the first round of massive changes, where
>> most
>> > of the internals are using Stream API instead of the Iterable system
>> from
>> > Rickard.
>> > It compiles and some non-trivial tests pass.
>> >
>> > BUT, I must have messed up some of the functions, because withTags()
>> > doesn't work anymore. When the withTags() are to be converted into a
>> > ServiceTags instance and placed into MetaInfo, something goes wrong and
>> > ServiceTags are never added. This breaks a LOT of tests, due to "json"
>> > tagging on ValueSerialization, and I haven't stumbled upon breaking for
>> > other reasons, so fixing this could(!) clear up a massive number of
>> tests
>> > in one go.
>> >
>> > I also note that the Build Passes, even though most tests fail. That
>> can't
>> > be right...
>> >
>> > I am committing and pushing this anyway, in case someone has time to
>> look
>> > at this detail tonight or tomorrow, as I need to run around with other
>> > things quite a lot, for my move to Shanghai.
>> >
>> > So, my guess is that CI will break and start making noise.
>> >
>>
>>
>
>
> --
> Niclas Hedhman, Software Developer
> http://zest.apache.org - New Energy for Java
>
--
Niclas Hedhman, Software Developer
http://zest.apache.org - New Energy for Java
Re: Stream/Functional progress report
Posted by Niclas Hedhman <ni...@hedhman.org>.
Yeah, I kind of figured that much out already. But I still can't spot the
difference from the old code.
And not comfortable of just adding it.
On Thu, Aug 27, 2015 at 6:53 AM, Kent Sølvsten <ke...@gmail.com>
wrote:
> Debugger tells that the bug is inside Classes#TYPES_OF.
>
> The old implementation returned both interfaces and parent classes for a
> class, - the new implementation returns only interfaces.
> The effect is that we only store the ServiceTags under metaInfo keyed by
> the Serializable interface, and thus are unable to find it.
>
> Who will be first to fix it?
>
> /Kent
>
> Den 26-08-2015 kl. 17:33 skrev Niclas Hedhman:
> > Gang,
> >
> > I want to give you an update on how it is going...
> >
> > In essence, I have completed the first round of massive changes, where
> most
> > of the internals are using Stream API instead of the Iterable system from
> > Rickard.
> > It compiles and some non-trivial tests pass.
> >
> > BUT, I must have messed up some of the functions, because withTags()
> > doesn't work anymore. When the withTags() are to be converted into a
> > ServiceTags instance and placed into MetaInfo, something goes wrong and
> > ServiceTags are never added. This breaks a LOT of tests, due to "json"
> > tagging on ValueSerialization, and I haven't stumbled upon breaking for
> > other reasons, so fixing this could(!) clear up a massive number of tests
> > in one go.
> >
> > I also note that the Build Passes, even though most tests fail. That
> can't
> > be right...
> >
> > I am committing and pushing this anyway, in case someone has time to look
> > at this detail tonight or tomorrow, as I need to run around with other
> > things quite a lot, for my move to Shanghai.
> >
> > So, my guess is that CI will break and start making noise.
> >
>
>
--
Niclas Hedhman, Software Developer
http://zest.apache.org - New Energy for Java
Re: Stream/Functional progress report
Posted by Kent Sølvsten <ke...@gmail.com>.
Debugger tells that the bug is inside Classes#TYPES_OF.
The old implementation returned both interfaces and parent classes for a
class, - the new implementation returns only interfaces.
The effect is that we only store the ServiceTags under metaInfo keyed by
the Serializable interface, and thus are unable to find it.
Who will be first to fix it?
/Kent
Den 26-08-2015 kl. 17:33 skrev Niclas Hedhman:
> Gang,
>
> I want to give you an update on how it is going...
>
> In essence, I have completed the first round of massive changes, where most
> of the internals are using Stream API instead of the Iterable system from
> Rickard.
> It compiles and some non-trivial tests pass.
>
> BUT, I must have messed up some of the functions, because withTags()
> doesn't work anymore. When the withTags() are to be converted into a
> ServiceTags instance and placed into MetaInfo, something goes wrong and
> ServiceTags are never added. This breaks a LOT of tests, due to "json"
> tagging on ValueSerialization, and I haven't stumbled upon breaking for
> other reasons, so fixing this could(!) clear up a massive number of tests
> in one go.
>
> I also note that the Build Passes, even though most tests fail. That can't
> be right...
>
> I am committing and pushing this anyway, in case someone has time to look
> at this detail tonight or tomorrow, as I need to run around with other
> things quite a lot, for my move to Shanghai.
>
> So, my guess is that CI will break and start making noise.
>