Re: Spatial model in MADlib (GWR)

[Frank McQuillan]

I created a JIRA for GWR
https://issues.apache.org/jira/browse/MADLIB-1023
and marked it for v1.9.2 for now though we can of course change that
depending on progress.





On Wed, Sep 14, 2016 at 9:36 AM, Rahul Iyer  wrote:

> Yes, you'll have to compile MADlib with DEBUG flag. Use `cmake
> -DCMAKE_BUILD_TYPE=Debug ..` to get MADlib symbols.
> You would benefit by compiling Postgres (with debug) as well but may not be
> necessary for what you're doing.
>
> On Tue, Sep 13, 2016 at 10:20 PM, Wang ChenLiang 
> wrote:
>
> > Hi Rahul
> >
> > Thanks for your reply. Should I build MADlib and Postgres with
> > enable-debug and no optimization flags? How can I list source code of
> > MADlib in GDB ? Sorry for asking such silly questions.
> >
> > Best,
> > Chenliang Wang
> >
> > On 09/14/2016 02:18 AM, Rahul Iyer wrote:
> > > Hi Chengliang
> > >
> > > There's some information on debugging
> > > ​in ​
> > > our old wiki page
> > >  > from-Source#debugging>
> > > ​. There's no example there but the process is simple once you have the
> > > server process id.
> > >
> > > - Rahul ​
> > >
> > > On Tue, Sep 13, 2016 at 6:41 AM, Wang ChenLiang 
> > wrote:
> > >
> > >> Hi Frank,
> > >>
> > >> I was being on a business trip for several months and began to work on
> > >> MADlib again in the past few days. But I have a trouble with debugging
> > >> MADlib with GDB. Could you kindly give me a detailed example for
> > >> debugging MADlib with CodeBlocks or GDB?
> > >>
> > >> Many Thanks !
> > >>
> > >>
> > >> On 03/15/2016 12:31 AM, Frank McQuillan wrote:
> > >>> OK.  Please don't hesitate to ask if you have any questions.
> > >>>
> > >>> Frank
> > >>>
> > >>> On Mon, Mar 14, 2016 at 4:17 AM, chenliang wang  >
> > >> wrote:
> > >>>
> >  Hi, Frank
> > 
> >  Recently,I am just looking at the detail of development guide and
> > trying
> >  to complete the serial algorithm. And I plan to implement GWR
> dividing
> >  the loop into pieces of chunks executed in several nodes. However, I
> > am
> >  not sure if there are some specials details need to be designed for
> >  distributed models in GPDB because I haven't developed model in MPP
> >  architecture. I hope this distributed manner would be implemented
> > >> easily.
> > 
> >  Best,
> >  Chenliang Wang
> > 
> >  On 03/10/2016 08:33 AM, Frank McQuillan wrote:
> > > Hi ChenLiang Wang,
> > >
> > > I am checking to see how things are going regarding the GWR model
> for
> > > MADlib that you proposed.  Not sure which phase you are at, but a
> >  suggested
> > > next step might be how you plan to implement the GWR algorithm in a
> > > distributed manner.  That is, how will it run in parallel?
> > >
> > > (Starting as a new thread since the previous thread fragmented.)
> > >
> > > Regards,
> > > Frank
> > >
> > 
> > >>>
> > >>
> > >
> >
>

Re: Spatial model in MADlib (GWR)

[Rahul Iyer]

Yes, you'll have to compile MADlib with DEBUG flag. Use `cmake
-DCMAKE_BUILD_TYPE=Debug ..` to get MADlib symbols.
You would benefit by compiling Postgres (with debug) as well but may not be
necessary for what you're doing.

On Tue, Sep 13, 2016 at 10:20 PM, Wang ChenLiang 
wrote:

> Hi Rahul
>
> Thanks for your reply. Should I build MADlib and Postgres with
> enable-debug and no optimization flags? How can I list source code of
> MADlib in GDB ? Sorry for asking such silly questions.
>
> Best,
> Chenliang Wang
>
> On 09/14/2016 02:18 AM, Rahul Iyer wrote:
> > Hi Chengliang
> >
> > There's some information on debugging
> > ​in ​
> > our old wiki page
> >  from-Source#debugging>
> > ​. There's no example there but the process is simple once you have the
> > server process id.
> >
> > - Rahul ​
> >
> > On Tue, Sep 13, 2016 at 6:41 AM, Wang ChenLiang 
> wrote:
> >
> >> Hi Frank,
> >>
> >> I was being on a business trip for several months and began to work on
> >> MADlib again in the past few days. But I have a trouble with debugging
> >> MADlib with GDB. Could you kindly give me a detailed example for
> >> debugging MADlib with CodeBlocks or GDB?
> >>
> >> Many Thanks !
> >>
> >>
> >> On 03/15/2016 12:31 AM, Frank McQuillan wrote:
> >>> OK.  Please don't hesitate to ask if you have any questions.
> >>>
> >>> Frank
> >>>
> >>> On Mon, Mar 14, 2016 at 4:17 AM, chenliang wang 
> >> wrote:
> >>>
>  Hi, Frank
> 
>  Recently,I am just looking at the detail of development guide and
> trying
>  to complete the serial algorithm. And I plan to implement GWR dividing
>  the loop into pieces of chunks executed in several nodes. However, I
> am
>  not sure if there are some specials details need to be designed for
>  distributed models in GPDB because I haven't developed model in MPP
>  architecture. I hope this distributed manner would be implemented
> >> easily.
> 
>  Best,
>  Chenliang Wang
> 
>  On 03/10/2016 08:33 AM, Frank McQuillan wrote:
> > Hi ChenLiang Wang,
> >
> > I am checking to see how things are going regarding the GWR model for
> > MADlib that you proposed.  Not sure which phase you are at, but a
>  suggested
> > next step might be how you plan to implement the GWR algorithm in a
> > distributed manner.  That is, how will it run in parallel?
> >
> > (Starting as a new thread since the previous thread fragmented.)
> >
> > Regards,
> > Frank
> >
> 
> >>>
> >>
> >
>

Re: Spatial model in MADlib (GWR)

[Wang ChenLiang]

Hi Rahul

Thanks for your reply. Should I build MADlib and Postgres with
enable-debug and no optimization flags? How can I list source code of
MADlib in GDB ? Sorry for asking such silly questions.

Best,
Chenliang Wang

On 09/14/2016 02:18 AM, Rahul Iyer wrote:
> Hi Chengliang
> 
> There's some information on debugging
> ​in ​
> our old wiki page
> 
> ​. There's no example there but the process is simple once you have the
> server process id.
> 
> - Rahul ​
> 
> On Tue, Sep 13, 2016 at 6:41 AM, Wang ChenLiang  wrote:
> 
>> Hi Frank,
>>
>> I was being on a business trip for several months and began to work on
>> MADlib again in the past few days. But I have a trouble with debugging
>> MADlib with GDB. Could you kindly give me a detailed example for
>> debugging MADlib with CodeBlocks or GDB?
>>
>> Many Thanks !
>>
>>
>> On 03/15/2016 12:31 AM, Frank McQuillan wrote:
>>> OK.  Please don't hesitate to ask if you have any questions.
>>>
>>> Frank
>>>
>>> On Mon, Mar 14, 2016 at 4:17 AM, chenliang wang 
>> wrote:
>>>
 Hi, Frank

 Recently,I am just looking at the detail of development guide and trying
 to complete the serial algorithm. And I plan to implement GWR dividing
 the loop into pieces of chunks executed in several nodes. However, I am
 not sure if there are some specials details need to be designed for
 distributed models in GPDB because I haven't developed model in MPP
 architecture. I hope this distributed manner would be implemented
>> easily.

 Best,
 Chenliang Wang

 On 03/10/2016 08:33 AM, Frank McQuillan wrote:
> Hi ChenLiang Wang,
>
> I am checking to see how things are going regarding the GWR model for
> MADlib that you proposed.  Not sure which phase you are at, but a
 suggested
> next step might be how you plan to implement the GWR algorithm in a
> distributed manner.  That is, how will it run in parallel?
>
> (Starting as a new thread since the previous thread fragmented.)
>
> Regards,
> Frank
>

>>>
>>
> 

Re: Spatial model in MADlib (GWR)

[Rahul Iyer]

Hi Chengliang

There's some information on debugging
​in ​
our old wiki page

​. There's no example there but the process is simple once you have the
server process id.

- Rahul ​

On Tue, Sep 13, 2016 at 6:41 AM, Wang ChenLiang  wrote:

> Hi Frank,
>
> I was being on a business trip for several months and began to work on
> MADlib again in the past few days. But I have a trouble with debugging
> MADlib with GDB. Could you kindly give me a detailed example for
> debugging MADlib with CodeBlocks or GDB?
>
> Many Thanks !
>
>
> On 03/15/2016 12:31 AM, Frank McQuillan wrote:
> > OK.  Please don't hesitate to ask if you have any questions.
> >
> > Frank
> >
> > On Mon, Mar 14, 2016 at 4:17 AM, chenliang wang 
> wrote:
> >
> >> Hi, Frank
> >>
> >> Recently,I am just looking at the detail of development guide and trying
> >> to complete the serial algorithm. And I plan to implement GWR dividing
> >> the loop into pieces of chunks executed in several nodes. However, I am
> >> not sure if there are some specials details need to be designed for
> >> distributed models in GPDB because I haven't developed model in MPP
> >> architecture. I hope this distributed manner would be implemented
> easily.
> >>
> >> Best,
> >> Chenliang Wang
> >>
> >> On 03/10/2016 08:33 AM, Frank McQuillan wrote:
> >>> Hi ChenLiang Wang,
> >>>
> >>> I am checking to see how things are going regarding the GWR model for
> >>> MADlib that you proposed.  Not sure which phase you are at, but a
> >> suggested
> >>> next step might be how you plan to implement the GWR algorithm in a
> >>> distributed manner.  That is, how will it run in parallel?
> >>>
> >>> (Starting as a new thread since the previous thread fragmented.)
> >>>
> >>> Regards,
> >>> Frank
> >>>
> >>
> >
>

Re: Spatial model in MADlib (GWR)

[Wang ChenLiang]

Great! I would like to join it.  Thanks!



From: Greg Chase 
Sent: Tuesday, September 13, 2016 21:55
To: dev@madlib.incubator.apache.org
Subject: Re: Spatial model in MADlib (GWR)

Hi ChenLang,
We have a community meeting at 9AM pacific. If you are still awake please feel 
free to ask the question in the end of the meeting.

Dial in instructions are in the meeting annoucemt and I will reply here again 
with them later.

Greg

This email encrypted by tiny buttons & fat thumbs, beta voice recognition, and 
autocorrect on my iPhone.

> On Sep 13, 2016, at 6:41 AM, Wang ChenLiang  wrote:
>
> Hi Frank,
>
> I was being on a business trip for several months and began to work on
> MADlib again in the past few days. But I have a trouble with debugging
> MADlib with GDB. Could you kindly give me a detailed example for
> debugging MADlib with CodeBlocks or GDB?
>
> Many Thanks !
>
>
>> On 03/15/2016 12:31 AM, Frank McQuillan wrote:
>> OK.  Please don't hesitate to ask if you have any questions.
>>
>> Frank
>>
>>> On Mon, Mar 14, 2016 at 4:17 AM, chenliang wang  wrote:
>>>
>>> Hi, Frank
>>>
>>> Recently,I am just looking at the detail of development guide and trying
>>> to complete the serial algorithm. And I plan to implement GWR dividing
>>> the loop into pieces of chunks executed in several nodes. However, I am
>>> not sure if there are some specials details need to be designed for
>>> distributed models in GPDB because I haven't developed model in MPP
>>> architecture. I hope this distributed manner would be implemented easily.
>>>
>>> Best,
>>> Chenliang Wang
>>>
>>>> On 03/10/2016 08:33 AM, Frank McQuillan wrote:
>>>> Hi ChenLiang Wang,
>>>>
>>>> I am checking to see how things are going regarding the GWR model for
>>>> MADlib that you proposed.  Not sure which phase you are at, but a
>>> suggested
>>>> next step might be how you plan to implement the GWR algorithm in a
>>>> distributed manner.  That is, how will it run in parallel?
>>>>
>>>> (Starting as a new thread since the previous thread fragmented.)
>>>>
>>>> Regards,
>>>> Frank
>>

Re: Spatial model in MADlib (GWR)

[Greg Chase]

Hi ChenLang, 
We have a community meeting at 9AM pacific. If you are still awake please feel 
free to ask the question in the end of the meeting.

Dial in instructions are in the meeting annoucemt and I will reply here again 
with them later.  

Greg 

This email encrypted by tiny buttons & fat thumbs, beta voice recognition, and 
autocorrect on my iPhone.

> On Sep 13, 2016, at 6:41 AM, Wang ChenLiang  wrote:
> 
> Hi Frank,
> 
> I was being on a business trip for several months and began to work on
> MADlib again in the past few days. But I have a trouble with debugging
> MADlib with GDB. Could you kindly give me a detailed example for
> debugging MADlib with CodeBlocks or GDB?
> 
> Many Thanks !
> 
> 
>> On 03/15/2016 12:31 AM, Frank McQuillan wrote:
>> OK.  Please don't hesitate to ask if you have any questions.
>> 
>> Frank
>> 
>>> On Mon, Mar 14, 2016 at 4:17 AM, chenliang wang  wrote:
>>> 
>>> Hi, Frank
>>> 
>>> Recently,I am just looking at the detail of development guide and trying
>>> to complete the serial algorithm. And I plan to implement GWR dividing
>>> the loop into pieces of chunks executed in several nodes. However, I am
>>> not sure if there are some specials details need to be designed for
>>> distributed models in GPDB because I haven't developed model in MPP
>>> architecture. I hope this distributed manner would be implemented easily.
>>> 
>>> Best,
>>> Chenliang Wang
>>> 
 On 03/10/2016 08:33 AM, Frank McQuillan wrote:
 Hi ChenLiang Wang,
 
 I am checking to see how things are going regarding the GWR model for
 MADlib that you proposed.  Not sure which phase you are at, but a
>>> suggested
 next step might be how you plan to implement the GWR algorithm in a
 distributed manner.  That is, how will it run in parallel?
 
 (Starting as a new thread since the previous thread fragmented.)
 
 Regards,
 Frank
>> 

Re: Spatial model in MADlib (GWR)

[Wang ChenLiang]

Hi Frank,

I was being on a business trip for several months and began to work on
MADlib again in the past few days. But I have a trouble with debugging
MADlib with GDB. Could you kindly give me a detailed example for
debugging MADlib with CodeBlocks or GDB?

Many Thanks !


On 03/15/2016 12:31 AM, Frank McQuillan wrote:
> OK.  Please don't hesitate to ask if you have any questions.
> 
> Frank
> 
> On Mon, Mar 14, 2016 at 4:17 AM, chenliang wang  wrote:
> 
>> Hi, Frank
>>
>> Recently,I am just looking at the detail of development guide and trying
>> to complete the serial algorithm. And I plan to implement GWR dividing
>> the loop into pieces of chunks executed in several nodes. However, I am
>> not sure if there are some specials details need to be designed for
>> distributed models in GPDB because I haven't developed model in MPP
>> architecture. I hope this distributed manner would be implemented easily.
>>
>> Best,
>> Chenliang Wang
>>
>> On 03/10/2016 08:33 AM, Frank McQuillan wrote:
>>> Hi ChenLiang Wang,
>>>
>>> I am checking to see how things are going regarding the GWR model for
>>> MADlib that you proposed.  Not sure which phase you are at, but a
>> suggested
>>> next step might be how you plan to implement the GWR algorithm in a
>>> distributed manner.  That is, how will it run in parallel?
>>>
>>> (Starting as a new thread since the previous thread fragmented.)
>>>
>>> Regards,
>>> Frank
>>>
>>
> 

Re: Spatial model in MADlib (GWR)

[Frank McQuillan]

OK.  Please don't hesitate to ask if you have any questions.

Frank

On Mon, Mar 14, 2016 at 4:17 AM, chenliang wang  wrote:

> Hi, Frank
>
> Recently,I am just looking at the detail of development guide and trying
> to complete the serial algorithm. And I plan to implement GWR dividing
> the loop into pieces of chunks executed in several nodes. However, I am
> not sure if there are some specials details need to be designed for
> distributed models in GPDB because I haven't developed model in MPP
> architecture. I hope this distributed manner would be implemented easily.
>
> Best,
> Chenliang Wang
>
> On 03/10/2016 08:33 AM, Frank McQuillan wrote:
> > Hi ChenLiang Wang,
> >
> > I am checking to see how things are going regarding the GWR model for
> > MADlib that you proposed.  Not sure which phase you are at, but a
> suggested
> > next step might be how you plan to implement the GWR algorithm in a
> > distributed manner.  That is, how will it run in parallel?
> >
> > (Starting as a new thread since the previous thread fragmented.)
> >
> > Regards,
> > Frank
> >
>

Re: Spatial model in MADlib (GWR)

[chenliang wang]

Hi, Frank

Recently,I am just looking at the detail of development guide and trying
to complete the serial algorithm. And I plan to implement GWR dividing
the loop into pieces of chunks executed in several nodes. However, I am
not sure if there are some specials details need to be designed for
distributed models in GPDB because I haven't developed model in MPP
architecture. I hope this distributed manner would be implemented easily.

Best,
Chenliang Wang

On 03/10/2016 08:33 AM, Frank McQuillan wrote:
> Hi ChenLiang Wang,
> 
> I am checking to see how things are going regarding the GWR model for
> MADlib that you proposed.  Not sure which phase you are at, but a suggested
> next step might be how you plan to implement the GWR algorithm in a
> distributed manner.  That is, how will it run in parallel?
> 
> (Starting as a new thread since the previous thread fragmented.)
> 
> Regards,
> Frank
> 

Spatial model in MADlib (GWR)

[Frank McQuillan]

Hi ChenLiang Wang,

I am checking to see how things are going regarding the GWR model for
MADlib that you proposed.  Not sure which phase you are at, but a suggested
next step might be how you plan to implement the GWR algorithm in a
distributed manner.  That is, how will it run in parallel?

(Starting as a new thread since the previous thread fragmented.)

Regards,
Frank