​Re: ​SCP Workflow question

Andy Curtis abcurtis at gmail.com
Wed May 9 06:13:43 EDT 2018


Maybe I am doing a poor job of explaining because we do have a team of
experienced mobile people.

We originally said, 'when user presses 'submit' there should be 2 things
happen, the 'request' should be saved and Wf should be triggered.  They
read this as 2 client side actions (1 to send screen data to the server,
the second to trigger Wf) and were concerned what would happen if the first
action succeeded and the second failed.  The answer to that is to have 1
server side service to receive the screen data, create a database record
and trigger Wf).  But the pushback came that if there was 1 client side
action to trigger Wf (server side) passing the screen data to the Context,
then Wf could create the database record as the first step.

Thats where my karma is suffering.  I believe it is 'right/correct' to have
a server side service receive the screen data, create a 'request' on a
database and then trigger Wf.  Replacing the server side service to receive
the screendata and trigger Wf, I think is fundamentally wrong because I
have never created a Wf that did not have an business object/Class instance
when it started.

Thoughts?

Andy


On Wed, May 9, 2018 at 10:11 AM, <sap-wug-request at mit.edu> wrote:

> Send SAP-WUG mailing list submissions to
>         sap-wug at mit.edu
>
> To subscribe or unsubscribe via the World Wide Web, visit
>         http://mailman.mit.edu/mailman/listinfo/sap-wug
> or, via email, send a message with subject or body 'help' to
>         sap-wug-request at mit.edu
>
> You can reach the person managing the list at
>         sap-wug-owner at mit.edu
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of SAP-WUG digest..."
>
> Today's Topics:
>
>    1. Re: ?SCP Workflow question (Dart, Jocelyn)
>
>
> ---------- Forwarded message ----------
> From: "Dart, Jocelyn" <jocelyn.dart at sap.com>
> To: "SAP Workflow Users' Group" <sap-wug at mit.edu>
> Cc:
> Bcc:
> Date: Wed, 9 May 2018 09:11:13 +0000
> Subject:
> ​​
> Re: ​SCP Workflow question
> Hi Andy
> Ok this sounds like the usual non workflow developer mistake of thinking
> workflow is somehow another pseudo database. With this approach if you lose
> comms you have an orphaned workflow.... which surely would be much harder
> to clean up than a single database entry wherever it is held.
>
> Do they have anyone who has actually done offline mobile scenarios before
> advising them?
>
> To me - and granted I am not an expert in this - it sounds more like what
> you need is a staging table in the cloud database for the initial request.
>   Then if all details are received that becomes your draft business object
> trigger for the workflow which then completes the process including
> identifying if the entry is a duplicate or receiving out of order or
> whatever & if all is well then updating the real backend database & then
> ends with cleaning up the draft/staging entry.
>
> That way you know if there is anything outstanding or blocked by checking
> the staging table contents
>
> Just a suggestion... again really they need to involve someone with
> offline mobile experience
> Rgds
> Jocelyn
>
> Sent from my iPhone with many apologies for the spelling, grammar and any
> other deficiencies
>
> On 9 May 2018, at 6:43 pm, Andy Curtis <abcurtis at gmail.com> wrote:
>
> Thanks WUG'ers
>
> ​I'll add some 'colour' to my scenario.  We have a UI5 screen to create a
> 'request' this screen can be delivered to a user on any device, including
> mobile phones where comms could be patchy, so loss of comms is a feature
> that the 'techs' are concerned about.  So the proposal from them is to have
> a single comm from the UI5 when the 'request' data is entered.  So, data
> coming from a screen on a mobile, arrives at the server and something needs
> to take it and create a 'request' in a database.  My opinion is there
> should be a Service to receive the comms and create the 'request' in the
> database, assigning a number to it and then triggering Wf passing the
> 'request' that has been saved on a database.
>
> Their suggestion is, when the data arrives at the server it triggers Wf
> immediately (before the 'request' is created on the database) and the first
> step in the Wf is to create the 'request' get back the 'request' nbr and
> then continue through the approval process.  This would be the first time I
> have ever started a Wf without an instance of the object, (I don't think I
> could trigger a Classic Wf without an object instance)​ and it just does
> not feel right to me.  Hence my question to the community to find out if
> others thought it was good practice to start a Wf without an object
> instance and have Wf create it.
>
>
> ​@Kjetil, in a create vendor account process, would the first step be a
> user in XK01 creating a Vendor account?  When saving the transaction it
> creates a vendor record and triggers Wf to start the process.  The vendor
> nbr would exist before Wf was triggered and an instance of the vendor
> object would exist.  Or maybe you have a Form and you want to start a Wf to
> create a vendor master  ​
>
> ​@Mark, I started to think SAP Cloud was the new R/3 and prepared to catch
> a new wave,   ​
>
>
>
>
> Andy Curtis
>
>
> On Wed, May 9, 2018 at 12:55 AM, <sap-wug-request at mit.edu> wrote:
>
>> Send SAP-WUG mailing list submissions to
>>         sap-wug at mit.edu
>>
>> To subscribe or unsubscribe via the World Wide Web, visit
>>         http://mailman.mit.edu/mailman/listinfo/sap-wug
>> or, via email, send a message with subject or body 'help' to
>>         sap-wug-request at mit.edu
>>
>> You can reach the person managing the list at
>>         sap-wug-owner at mit.edu
>>
>> When replying, please edit your Subject line so it is more specific
>> than "Re: Contents of SAP-WUG digest..."
>>
>> Today's Topics:
>>
>>    1. SCP Workflow question (Andy Curtis)
>>    2. Re: SCP Workflow question (Kjetil Kilhavn)
>>    3. Re: SCP Workflow question (Mark Pyc)
>>
>>
>> ---------- Forwarded message ----------
>> From: Andy Curtis <abcurtis at gmail.com>
>> To: "SAP Workflow Users' Group" <sap-wug at mit.edu>
>> Cc:
>> Bcc:
>> Date: Tue, 8 May 2018 18:08:44 +0100
>> Subject:
>> ​​
>> SCP Workflow question
>> WUG'ers :) (longtime no see)
>>
>> I am graduating to SCP Workflow and have a question, I wonder if anyone
>> can help me out.
>>
>> I have always built Classic Workflows triggered after an object has been
>> created in a database, or SAP always saves the object and then sends the Wf
>> triggering event.  I would say thats best practice and it also fits with my
>> other narrative about keeping process and application logic separate.
>>
>> In SCP it is possible to trigger a SCP Workflow without an Object, but is
>> that a good idea?  The thought it to trigger SCP Wf and have a step that
>> calls a task to create the object, so the Wf is basically generic to start
>> with, then becomes instantiated after the first step.  Would anyone else
>> think this is a good idea?  I don't, I think a Wf should have an object
>> before being started but I am having a hard time arguing the case, so
>> really looking for other informed opinion.
>>
>> Anyone got one?
>>
>>
>> Andy Curtis
>>
>>
>>
>> ---------- Forwarded message ----------
>> From: Kjetil Kilhavn <list.sap-wug at vettug.no>
>> To: sap-wug at mit.edu
>> Cc:
>> Bcc:
>> Date: Tue, 8 May 2018 22:03:13 +0200
>> Subject: Re: SCP Workflow question
>>
>> What is an object? If the workflow is "Create Vendor Account" and takes a
>> bunch of input parameters for the new vendor, then why not let it start
>> before there is an LFA1 record in the database? You have an object, it just
>> isn't complete yet, it is a draft for a vendor record. Sort of like FIPP
>> objects.
>>
>> But the way you describe it it may also be a workflow that doesn't have a
>> clue what is about to happen, so any type of object can be created etc. I
>> suppose it could still be a valid use case, but it's a bit harder to
>> imagine. How would one for instance identify agents for the first step if
>> you don't even know what is going to be created.
>>
>> Den 08. mai 2018 19:08, skreiv Andy Curtis:
>>
>> WUG'ers :) (longtime no see)
>>
>> I am graduating to SCP Workflow and have a question, I wonder if anyone
>> can help me out.
>>
>> I have always built Classic Workflows triggered after an object has been
>> created in a database, or SAP always saves the object and then sends the Wf
>> triggering event.  I would say thats best practice and it also fits with my
>> other narrative about keeping process and application logic separate.
>>
>> In SCP it is possible to trigger a SCP Workflow without an Object, but is
>> that a good idea?  The thought it to trigger SCP Wf and have a step that
>> calls a task to create the object, so the Wf is basically generic to start
>> with, then becomes instantiated after the first step.  Would anyone else
>> think this is a good idea?  I don't, I think a Wf should have an object
>> before being started but I am having a hard time arguing the case, so
>> really looking for other informed opinion.
>>
>> Anyone got one?
>>
>>
>> Andy Curtis
>>
>>
>>
>> _______________________________________________
>> SAP-WUG mailing listSAP-WUG at mit.eduhttp://mailman.mit.edu/mailman/listinfo/sap-wug
>>
>>
>> --
>> Kjetil Kilhavn
>>
>>
>>
>> ---------- Forwarded message ----------
>> From: Mark Pyc <mark.pyc at gmail.com>
>> To: "SAP Workflow Users' Group" <sap-wug at mit.edu>
>> Cc:
>> Bcc:
>> Date: Wed, 9 May 2018 09:50:14 +1000
>> Subject: Re: SCP Workflow question
>> Agree with Kjetil and agree it's a lot like a FIPP to BKPF or maybe more
>> so like BUS2105 to BUS2012. Thing is LFA1 doesn't have such an obvious
>> 'requested' phase. I've done similar things before but I end up creating a
>> request application which assigns a GUID to the request, and that's the
>> object that the WF operates on. Now you _could_ use the WF instance id as
>> the request number and capture all the data into containers rather than
>> custom tables but, in non-cloud at least, that would seem to me to be a
>> lazy hack that would cost you in the long run - think about reporting and
>> monitoring.
>>
>> The cloud... hmmm don't get me started on the cloud, or least SAPs
>> endeavours in that space... it might seem easier to use the WF as the
>> request DB but again I still think it's lazy. I started typing thinking I
>> might have a different attitude given the cloud but I don't
>>
>> Create a "Vendor Request" application (if you don't want to use MDM
>> requests and WF - not sure of cloud availability) and then build WF with a
>> known instance as you always would.
>>
>> My 2p / 2c / 2 lowest units of your desired currency.
>>
>> Have fun,
>> Mark
>>
>> On 9 May 2018 at 06:03, Kjetil Kilhavn <list.sap-wug at vettug.no> wrote:
>>
>>> What is an object? If the workflow is "Create Vendor Account" and takes
>>> a bunch of input parameters for the new vendor, then why not let it start
>>> before there is an LFA1 record in the database? You have an object, it just
>>> isn't complete yet, it is a draft for a vendor record. Sort of like FIPP
>>> objects.
>>>
>>> But the way you describe it it may also be a workflow that doesn't have
>>> a clue what is about to happen, so any type of object can be created etc. I
>>> suppose it could still be a valid use case, but it's a bit harder to
>>> imagine. How would one for instance identify agents for the first step if
>>> you don't even know what is going to be created.
>>>
>>> Den 08. mai 2018 19:08, skreiv Andy Curtis:
>>>
>>> WUG'ers :) (longtime no see)
>>>
>>> I am graduating to SCP Workflow and have a question, I wonder if anyone
>>> can help me out.
>>>
>>> I have always built Classic Workflows triggered after an object has been
>>> created in a database, or SAP always saves the object and then sends the Wf
>>> triggering event.  I would say thats best practice and it also fits with my
>>> other narrative about keeping process and application logic separate.
>>>
>>> In SCP it is possible to trigger a SCP Workflow without an Object, but
>>> is that a good idea?  The thought it to trigger SCP Wf and have a step that
>>> calls a task to create the object, so the Wf is basically generic to start
>>> with, then becomes instantiated after the first step.  Would anyone else
>>> think this is a good idea?  I don't, I think a Wf should have an object
>>> before being started but I am having a hard time arguing the case, so
>>> really looking for other informed opinion.
>>>
>>> Anyone got one?
>>>
>>>
>>> Andy Curtis
>>>
>>>
>>>
>>> _______________________________________________
>>> SAP-WUG mailing listSAP-WUG at mit.eduhttp://mailman.mit.edu/mailman/listinfo/sap-wug
>>>
>>>
>>> --
>>> Kjetil Kilhavn
>>>
>>>
>>> _______________________________________________
>>> SAP-WUG mailing list
>>> SAP-WUG at mit.edu
>>> http://mailman.mit.edu/mailman/listinfo/sap-wug
>>>
>>>
>>
>> _______________________________________________
>> SAP-WUG mailing list
>> SAP-WUG at mit.edu
>> http://mailman.mit.edu/mailman/listinfo/sap-wug
>>
>>
> _______________________________________________
> SAP-WUG mailing list
> SAP-WUG at mit.edu
> http://mailman.mit.edu/mailman/listinfo/sap-wug
>
>
> _______________________________________________
> SAP-WUG mailing list
> SAP-WUG at mit.edu
> http://mailman.mit.edu/mailman/listinfo/sap-wug
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/sap-wug/attachments/20180509/b4848631/attachment-0001.html


More information about the SAP-WUG mailing list