Ed

I agree this needs to be discussed but it is important to acknowledge that
THIS Fineract listserv is the only official (and required) discussion
space.  It is not the intention to bury anything in “an email thread”.

When I started up Mifos we spent a lot of time looking at Collection Sheets
and designing process flows around them. I fully know that this design
direction was important back then in 2002-2006.

However if there is no one here asking for them to be retained besides you,
then that is a sign that they have perhaps reached an end of their
utility.  Or, that the users are not actually here, which is a different
problem.

So,… If there is a group using collection sheets in production AND they are
not on some permanent forked (old) version, then now is the time to speak
up.  Here.

Generally, I’m fairly certain we can refactor this with an eye toward
extracting it from the core.  Repeating the logic in two places makes no
sense either.  Collection sheets are kind of assembled from
constituent loans and savings.  The balances and due payments should be
calculated in the underlying components.

As the system gets restructured we need to decide to keep this at all, to
keep it in a new place, or as some external concept/plug in. Why wouldn’t
we want a separate component?

Cheers

On Thu, Aug 14, 2025 at 4:16 PM Ed Cable <edca...@mifos.org> wrote:

> I'll leave the other thread for discussion of the API versioning and
> refactoring related to Collection Sheet API but wanted to create a separate
> thread regarding the deprecation of the Collection Sheet.
>
> In general, for this and removal of any functionality, it's something that
> needs to be discussed openly with the community and with a vote and not a
> decision buried in a mailing list thread.
>
> For the collection sheet specifically, more thought has to be given to its
> deprecation as the centrality and highly coupled nature of the collection
> sheet is being understated as it isn't merely a report that's to be printed
> or a form filled out via a mobile application. It's a significant portion
> of the user interface and highly coupled to many of the microfinance
> features around groups/centers/meeting scheduling, staff assignment, etc.
>
> I do agree that from a UI perspective, the collection sheet and other
> microfinance-centric functionalities and flows should be viewable based on
> a configurable setting. As it doesn't lend itself to the optimal user
> experience for a large portion of current Fineract user base.
>
> I also am supportive of a strategy of slimming Fineract to its core
> services and functionality above core Fineract services and APIs can be
> extracted out.
>
> So I do think we should give thoughtful consideration to what abstracting
> out the collection sheet and corresponding microfinance functionality would
> look like and what that effort would entail to abstract it out without
> adversely impacting the original user base of the software but it's not as
> simple as deprecating these API.
>
> I welcome others' thoughts and inputs as I know even with microfinance
> itself, the methodology has evolved and group lending and the concept of a
> collection sheet isn't as central as it once was
>
> Thanks,
>
> Ed
>
>
>
> On Wed, Aug 13, 2025 at 6:36 PM Kapil Panchal <
> kapil.panchal.developm...@gmail.com> wrote:
>
>> Hi James,
>>
>> I can proceed by marking this feature as @Deprecated and/or performing a
>> safe refactor to remove the API.
>>
>> For API versioning I agree to Aleksandar Vidakovic's proposal on adapting
>> to the SpringBoot v7/Spring Framework v4. If I may, Aleksandar
>> Vidakovic takes the lead on this project and I can help to support the
>> conversion?
>>
>> Thanks,
>> Kapil
>>
>> On Thu, Aug 14, 2025 at 5:07 AM James Dailey <jdai...@apache.org> wrote:
>>
>>> Hi Kapil
>>>
>>> I might suggest looking at this as an opportunity to remove the
>>> collection sheet entirely from the Fineract namespace.  It’s a legacy
>>> concept I and others designed a long time ago, originally in 2002 based on
>>> collection sheets we gathered from a dozen countries. It is strongly tied
>>> to concepts in microfinance field operations, and especially when there was
>>> no data connectivity.
>>>
>>> It belongs perhaps as a sort of external microservice - data loading via
>>> a bulk import could still be enabled.
>>>
>>> The API versioning is a good idea but needs to be more holistic across
>>> the platform I think.
>>>
>>>
>>>
>>> On Mon, Aug 11, 2025 at 4:43 AM Ádám Sághy <adamsa...@gmail.com> wrote:
>>>
>>>> Hi Kapil,
>>>>
>>>> Thank you for raising the concerns below. I’ll need some additional
>>>> details to fully understand your points:
>>>>
>>>>    1.
>>>>
>>>>    *Collection Sheet API* – You mentioned it appears non-functional
>>>>    and contains several logical errors.
>>>>    -
>>>>
>>>>       If it’s indeed not working, that’s a separate, high-priority
>>>>       discussion.
>>>>       -
>>>>
>>>>       Could you clarify which logical errors you were referring to,
>>>>       and what specifically makes you think it’s non-functional?
>>>>       2.
>>>>
>>>>    *Service annotations* – You noted that service methods are not
>>>>    annotated with @Service and that beans are defined manually.
>>>>    -
>>>>
>>>>       Are you referring to the
>>>>       CollectionSheetWritePlatformServiceJpaRepositoryImpl bean being
>>>>       defined via configuration?
>>>>       3.
>>>>
>>>>    *Repository wrappers annotated with @Service* – You mentioned that
>>>>    this mandates full unit test coverage but that they should ideally be
>>>>    annotated with @Component.
>>>>    -
>>>>
>>>>       Could you point out the exact classes you had in mind?
>>>>
>>>> As for the other points, I agree we can refactor and remove redundant
>>>> logic—please feel free to suggest specific improvements or start work on
>>>> them immediately!
>>>>
>>>> However, be careful by moving anything into the fineract-core… We are
>>>> aiming to keep it as small as possible as everything is built on top of
>>>> this module! If collection sheet are used for loans and savings - for
>>>> example - than the recommended move is NOT to move this logic into core!
>>>>
>>>> Either:
>>>>
>>>> - we split the logic into fineract-loan and fineract-savings
>>>>
>>>> - Move the logic into a new module
>>>>
>>>> - Leave it in fineract-provider for now
>>>>
>>>>
>>>> Shall you have any questions, please let us know!
>>>>
>>>> Regards,
>>>> Adam
>>>>
>>>>
>>>> On 2025. Aug 11., at 12:09, Kapil Panchal <
>>>> kapil.panchal.developm...@gmail.com> wrote:
>>>>
>>>> Hi Adam,
>>>>
>>>> I’m currently working on *FINERACT-2290* and have a few questions
>>>> before I submit a pull request.
>>>>
>>>> The *Collection Sheet API* in its current state appears non-functional
>>>> and contains several logical errors. It seems there was an earlier attempt
>>>> to convert from a JSON string request parameter to a class-based request
>>>> object, but:
>>>>
>>>> Certain fields are missing.
>>>>
>>>> The serializer is not correctly populating the objects, which causes
>>>> the conditional checks to be bypassed and results in incorrect (false)
>>>> responses.
>>>>
>>>> This change set is *high risk* because it touches most of the loan and
>>>> savings product logic. I’ve had to refactor almost all major methods.
>>>> Extensive integration and end-to-end testing will be required to ensure
>>>> there are no regressions, especially in edge cases. At present, there are
>>>> no unit or integration tests for this functionality, and test creation is
>>>> outside the current ticket scope. I’ve been iterating on this for a while,
>>>> and only today have I reached a stable state after several experimental and
>>>> build-breaking attempts.
>>>>
>>>> *Key Observations:*
>>>>
>>>> Service methods are not annotated with @Service; instead, beans are
>>>> defined manually.
>>>>
>>>> Repository wrappers are annotated with @Service. This mandates full
>>>> unit test coverage for these methods, but they should ideally be annotated
>>>> with @Component.
>>>>
>>>> I agree with prior discussions on separating bean validation — having a
>>>> dedicated @Component validation class allows the request object to handle
>>>> checks independent of database queries.
>>>>
>>>> Validation components can also perform database-related validations;
>>>> these can be injected into service classes for cleaner architecture.
>>>>
>>>> Such validation components should be placed in *Fineract-Core* so they
>>>> are reusable across modules, reducing future refactoring needs.
>>>>
>>>> The current design of having commands in *Fineract-Core* and
>>>> handlers/services/repositories in respective modules is good — it cleanly
>>>> decouples command definition from execution.
>>>>
>>>> There is extensive use of this. in singleton contexts (API, Service,
>>>> Repository). While not harmful, it’s unnecessary boilerplate.
>>>>
>>>> Multiple redundant intermediate DTOs exist where the request DTO itself
>>>> could be reused for data transfer.
>>>>
>>>> I found redundant logic — e.g., a for loop with a break statement that
>>>> effectively executes only once; this can be simplified.
>>>>
>>>> Some JDBC template queries use reserved SQL keywords, causing
>>>> exceptions. Refactoring these queries resolves the issue and returns proper
>>>> response objects.
>>>>
>>>> *Suggestions:*
>>>>
>>>> *Where appropriate, large tickets should be broken into subtasks to
>>>> manage complexity and reviewability.*
>>>>
>>>> It may help to have a dedicated *developer-only Slack channel *for
>>>> technical discussions. This could complement other community spaces if
>>>> there’s a need to keep certain conversations more focused.
>>>>
>>>> What are your thoughts on the above?
>>>>
>>>> Thanks,
>>>> Kapil
>>>>
>>>>
>>>>
>
> --
> *Ed Cable*
> President/CEO, Mifos Initiative
> edca...@mifos.org | Skype: edcable | Mobile: +1.484.477.8649
>
> *Collectively Creating a World of 3 Billion Maries | *http://mifos.org
> <http://facebook.com/mifos>  <http://www.twitter.com/mifos>
>
>

Reply via email to