You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ofbiz.apache.org by Konstantinos Marinos <co...@kmarinos.com> on 2026/07/28 19:59:21 UTC
[DISCUSS] Regarding export="true" for REST services
Hi all,
I was hoping to start a discussion about the current interaction of setting
export="true" in a service definition in regards to the recently added
rest-api module in the framework. The same flag that existed before
(export="true") now additionally exposes a service definition as a REST
endpoint. This is one of the ways to create a REST endpoint (the other
major one being a *.rest.xml definition file) but since the component is
now part of the framework, this could have unintended consequences.
My biggest concern is that developers might not immediately realise that
this one flag is used for similar but distinct use cases and this might not
be the desired behaviour for all new or previously exported services.
In order to avoid implicitly exposing services with potentially unintended
consequences, I am reaching out for your thoughts on the following
alternative actions:
* We remove the feature of defining REST endpoints in this matter
completely. Previously existing usages of export="true" remain unaffected
and REST endpoints can be defined by dedicated rest.xml files.
* We create a separate flag in the service definition (e.g.
export-api="true") that only controls the auto discovery and creation of
these REST endpoints. The two export features are then decoupled from each
other and by setting the new flag in the service definition, clear intent
is communicated by the developers.
What do you think about these two options?
If you are already using this feature to create REST endpoints or plan to
use it in the future, please let me know as well, as it would mean that the
second option has merit and it is the least destructive change of the two.
Thank you and best regards,
Konstantinos Marinos
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Konstantinos Marinos <co...@kmarinos.com>.
Hi all,
Thank you very much to everyone who participated in the discussion.
Since everyone agreed to one of the two options, I have prepared a PR that
implements the changes https://github.com/apache/ofbiz-framework/pull/1609
If anyone would like to review the changes, I would be very grateful.
Otherwise I am looking to merge the PR in the next few days.
Best regards,
Konstantinos Marinos
On Fri, Jul 31, 2026 at 6:40 AM Ratnesh Upadhyay <up...@gmail.com>
wrote:
> +1
>
> Thanks & Regards,
> Ratnesh Upadhyay
>
> On Thu, 30 Jul 2026 at 15:50, Deepak Dixit <de...@apache.org> wrote:
>
> > +1
> >
> > Thanks & Regards
> > --
> > Deepak Dixit
> > ofbiz.apache.org
> >
> >
> > On Thu, Jul 30, 2026 at 3:35 PM Jacopo Cappellato <
> > jacopo.cappellato@gmail.com> wrote:
> >
> > > As a side note, one advantage of removing the export=true feature from
> > > the rest-api plugin is that it will make the implementation of that
> > > plugin simpler (because this feature requires special handling).
> > >
> > > Jacopo
> > >
> > > On Thu, Jul 30, 2026 at 10:22 AM Michael Brohl <
> michael.brohl@ecomify.de
> > >
> > > wrote:
> > > >
> > > > +1
> > > >
> > > > Thanks,
> > > >
> > > > Michael Brohl
> > > >
> > > > ecomify GmbH - www.ecomify.de
> > > >
> > > >
> > > > Am 29.07.26 um 09:14 schrieb Mridul Pathak:
> > > > > I prefer the first approach. It's better to have a single way to
> > > expose any
> > > > > service as a REST endpoint through rest.xml. Those already using
> the
> > > > > current feature will need to migrate anyway.
> > > > >
> > > > > Thanks
> > > > > Mridul Pathak
> > > > >
> > > > > On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <
> > > contact@kmarinos.com>
> > > > > wrote:
> > > > >
> > > > >> Hi all,
> > > > >>
> > > > >> I was hoping to start a discussion about the current interaction
> of
> > > setting
> > > > >> export="true" in a service definition in regards to the recently
> > added
> > > > >> rest-api module in the framework. The same flag that existed
> before
> > > > >> (export="true") now additionally exposes a service definition as a
> > > REST
> > > > >> endpoint. This is one of the ways to create a REST endpoint (the
> > other
> > > > >> major one being a *.rest.xml definition file) but since the
> > component
> > > is
> > > > >> now part of the framework, this could have unintended
> consequences.
> > > > >>
> > > > >> My biggest concern is that developers might not immediately
> realise
> > > that
> > > > >> this one flag is used for similar but distinct use cases and this
> > > might not
> > > > >> be the desired behaviour for all new or previously exported
> > services.
> > > > >>
> > > > >> In order to avoid implicitly exposing services with potentially
> > > unintended
> > > > >> consequences, I am reaching out for your thoughts on the following
> > > > >> alternative actions:
> > > > >>
> > > > >> * We remove the feature of defining REST endpoints in this matter
> > > > >> completely. Previously existing usages of export="true" remain
> > > unaffected
> > > > >> and REST endpoints can be defined by dedicated rest.xml files.
> > > > >>
> > > > >> * We create a separate flag in the service definition (e.g.
> > > > >> export-api="true") that only controls the auto discovery and
> > creation
> > > of
> > > > >> these REST endpoints. The two export features are then decoupled
> > from
> > > each
> > > > >> other and by setting the new flag in the service definition, clear
> > > intent
> > > > >> is communicated by the developers.
> > > > >>
> > > > >> What do you think about these two options?
> > > > >>
> > > > >> If you are already using this feature to create REST endpoints or
> > > plan to
> > > > >> use it in the future, please let me know as well, as it would mean
> > > that the
> > > > >> second option has merit and it is the least destructive change of
> > the
> > > two.
> > > > >>
> > > > >> Thank you and best regards,
> > > > >> Konstantinos Marinos
> > > > >>
> > >
> >
>
>
> --
>
> Best Regards,
> Ratnesh Upadhyay
>
> *HotWax Systems*
> *Enterprise open source experts*
>
> http://www.hotwaxsystems.com
>
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Ratnesh Upadhyay <up...@gmail.com>.
+1
Thanks & Regards,
Ratnesh Upadhyay
On Thu, 30 Jul 2026 at 15:50, Deepak Dixit <de...@apache.org> wrote:
> +1
>
> Thanks & Regards
> --
> Deepak Dixit
> ofbiz.apache.org
>
>
> On Thu, Jul 30, 2026 at 3:35 PM Jacopo Cappellato <
> jacopo.cappellato@gmail.com> wrote:
>
> > As a side note, one advantage of removing the export=true feature from
> > the rest-api plugin is that it will make the implementation of that
> > plugin simpler (because this feature requires special handling).
> >
> > Jacopo
> >
> > On Thu, Jul 30, 2026 at 10:22 AM Michael Brohl <michael.brohl@ecomify.de
> >
> > wrote:
> > >
> > > +1
> > >
> > > Thanks,
> > >
> > > Michael Brohl
> > >
> > > ecomify GmbH - www.ecomify.de
> > >
> > >
> > > Am 29.07.26 um 09:14 schrieb Mridul Pathak:
> > > > I prefer the first approach. It's better to have a single way to
> > expose any
> > > > service as a REST endpoint through rest.xml. Those already using the
> > > > current feature will need to migrate anyway.
> > > >
> > > > Thanks
> > > > Mridul Pathak
> > > >
> > > > On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <
> > contact@kmarinos.com>
> > > > wrote:
> > > >
> > > >> Hi all,
> > > >>
> > > >> I was hoping to start a discussion about the current interaction of
> > setting
> > > >> export="true" in a service definition in regards to the recently
> added
> > > >> rest-api module in the framework. The same flag that existed before
> > > >> (export="true") now additionally exposes a service definition as a
> > REST
> > > >> endpoint. This is one of the ways to create a REST endpoint (the
> other
> > > >> major one being a *.rest.xml definition file) but since the
> component
> > is
> > > >> now part of the framework, this could have unintended consequences.
> > > >>
> > > >> My biggest concern is that developers might not immediately realise
> > that
> > > >> this one flag is used for similar but distinct use cases and this
> > might not
> > > >> be the desired behaviour for all new or previously exported
> services.
> > > >>
> > > >> In order to avoid implicitly exposing services with potentially
> > unintended
> > > >> consequences, I am reaching out for your thoughts on the following
> > > >> alternative actions:
> > > >>
> > > >> * We remove the feature of defining REST endpoints in this matter
> > > >> completely. Previously existing usages of export="true" remain
> > unaffected
> > > >> and REST endpoints can be defined by dedicated rest.xml files.
> > > >>
> > > >> * We create a separate flag in the service definition (e.g.
> > > >> export-api="true") that only controls the auto discovery and
> creation
> > of
> > > >> these REST endpoints. The two export features are then decoupled
> from
> > each
> > > >> other and by setting the new flag in the service definition, clear
> > intent
> > > >> is communicated by the developers.
> > > >>
> > > >> What do you think about these two options?
> > > >>
> > > >> If you are already using this feature to create REST endpoints or
> > plan to
> > > >> use it in the future, please let me know as well, as it would mean
> > that the
> > > >> second option has merit and it is the least destructive change of
> the
> > two.
> > > >>
> > > >> Thank you and best regards,
> > > >> Konstantinos Marinos
> > > >>
> >
>
--
Best Regards,
Ratnesh Upadhyay
*HotWax Systems*
*Enterprise open source experts*
http://www.hotwaxsystems.com
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Deepak Dixit <de...@apache.org>.
+1
Thanks & Regards
--
Deepak Dixit
ofbiz.apache.org
On Thu, Jul 30, 2026 at 3:35 PM Jacopo Cappellato <
jacopo.cappellato@gmail.com> wrote:
> As a side note, one advantage of removing the export=true feature from
> the rest-api plugin is that it will make the implementation of that
> plugin simpler (because this feature requires special handling).
>
> Jacopo
>
> On Thu, Jul 30, 2026 at 10:22 AM Michael Brohl <mi...@ecomify.de>
> wrote:
> >
> > +1
> >
> > Thanks,
> >
> > Michael Brohl
> >
> > ecomify GmbH - www.ecomify.de
> >
> >
> > Am 29.07.26 um 09:14 schrieb Mridul Pathak:
> > > I prefer the first approach. It's better to have a single way to
> expose any
> > > service as a REST endpoint through rest.xml. Those already using the
> > > current feature will need to migrate anyway.
> > >
> > > Thanks
> > > Mridul Pathak
> > >
> > > On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <
> contact@kmarinos.com>
> > > wrote:
> > >
> > >> Hi all,
> > >>
> > >> I was hoping to start a discussion about the current interaction of
> setting
> > >> export="true" in a service definition in regards to the recently added
> > >> rest-api module in the framework. The same flag that existed before
> > >> (export="true") now additionally exposes a service definition as a
> REST
> > >> endpoint. This is one of the ways to create a REST endpoint (the other
> > >> major one being a *.rest.xml definition file) but since the component
> is
> > >> now part of the framework, this could have unintended consequences.
> > >>
> > >> My biggest concern is that developers might not immediately realise
> that
> > >> this one flag is used for similar but distinct use cases and this
> might not
> > >> be the desired behaviour for all new or previously exported services.
> > >>
> > >> In order to avoid implicitly exposing services with potentially
> unintended
> > >> consequences, I am reaching out for your thoughts on the following
> > >> alternative actions:
> > >>
> > >> * We remove the feature of defining REST endpoints in this matter
> > >> completely. Previously existing usages of export="true" remain
> unaffected
> > >> and REST endpoints can be defined by dedicated rest.xml files.
> > >>
> > >> * We create a separate flag in the service definition (e.g.
> > >> export-api="true") that only controls the auto discovery and creation
> of
> > >> these REST endpoints. The two export features are then decoupled from
> each
> > >> other and by setting the new flag in the service definition, clear
> intent
> > >> is communicated by the developers.
> > >>
> > >> What do you think about these two options?
> > >>
> > >> If you are already using this feature to create REST endpoints or
> plan to
> > >> use it in the future, please let me know as well, as it would mean
> that the
> > >> second option has merit and it is the least destructive change of the
> two.
> > >>
> > >> Thank you and best regards,
> > >> Konstantinos Marinos
> > >>
>
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Jacopo Cappellato <ja...@gmail.com>.
As a side note, one advantage of removing the export=true feature from
the rest-api plugin is that it will make the implementation of that
plugin simpler (because this feature requires special handling).
Jacopo
On Thu, Jul 30, 2026 at 10:22 AM Michael Brohl <mi...@ecomify.de> wrote:
>
> +1
>
> Thanks,
>
> Michael Brohl
>
> ecomify GmbH - www.ecomify.de
>
>
> Am 29.07.26 um 09:14 schrieb Mridul Pathak:
> > I prefer the first approach. It's better to have a single way to expose any
> > service as a REST endpoint through rest.xml. Those already using the
> > current feature will need to migrate anyway.
> >
> > Thanks
> > Mridul Pathak
> >
> > On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <co...@kmarinos.com>
> > wrote:
> >
> >> Hi all,
> >>
> >> I was hoping to start a discussion about the current interaction of setting
> >> export="true" in a service definition in regards to the recently added
> >> rest-api module in the framework. The same flag that existed before
> >> (export="true") now additionally exposes a service definition as a REST
> >> endpoint. This is one of the ways to create a REST endpoint (the other
> >> major one being a *.rest.xml definition file) but since the component is
> >> now part of the framework, this could have unintended consequences.
> >>
> >> My biggest concern is that developers might not immediately realise that
> >> this one flag is used for similar but distinct use cases and this might not
> >> be the desired behaviour for all new or previously exported services.
> >>
> >> In order to avoid implicitly exposing services with potentially unintended
> >> consequences, I am reaching out for your thoughts on the following
> >> alternative actions:
> >>
> >> * We remove the feature of defining REST endpoints in this matter
> >> completely. Previously existing usages of export="true" remain unaffected
> >> and REST endpoints can be defined by dedicated rest.xml files.
> >>
> >> * We create a separate flag in the service definition (e.g.
> >> export-api="true") that only controls the auto discovery and creation of
> >> these REST endpoints. The two export features are then decoupled from each
> >> other and by setting the new flag in the service definition, clear intent
> >> is communicated by the developers.
> >>
> >> What do you think about these two options?
> >>
> >> If you are already using this feature to create REST endpoints or plan to
> >> use it in the future, please let me know as well, as it would mean that the
> >> second option has merit and it is the least destructive change of the two.
> >>
> >> Thank you and best regards,
> >> Konstantinos Marinos
> >>
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Michael Brohl <mi...@ecomify.de>.
+1
Thanks,
Michael Brohl
ecomify GmbH - www.ecomify.de
Am 29.07.26 um 09:14 schrieb Mridul Pathak:
> I prefer the first approach. It's better to have a single way to expose any
> service as a REST endpoint through rest.xml. Those already using the
> current feature will need to migrate anyway.
>
> Thanks
> Mridul Pathak
>
> On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <co...@kmarinos.com>
> wrote:
>
>> Hi all,
>>
>> I was hoping to start a discussion about the current interaction of setting
>> export="true" in a service definition in regards to the recently added
>> rest-api module in the framework. The same flag that existed before
>> (export="true") now additionally exposes a service definition as a REST
>> endpoint. This is one of the ways to create a REST endpoint (the other
>> major one being a *.rest.xml definition file) but since the component is
>> now part of the framework, this could have unintended consequences.
>>
>> My biggest concern is that developers might not immediately realise that
>> this one flag is used for similar but distinct use cases and this might not
>> be the desired behaviour for all new or previously exported services.
>>
>> In order to avoid implicitly exposing services with potentially unintended
>> consequences, I am reaching out for your thoughts on the following
>> alternative actions:
>>
>> * We remove the feature of defining REST endpoints in this matter
>> completely. Previously existing usages of export="true" remain unaffected
>> and REST endpoints can be defined by dedicated rest.xml files.
>>
>> * We create a separate flag in the service definition (e.g.
>> export-api="true") that only controls the auto discovery and creation of
>> these REST endpoints. The two export features are then decoupled from each
>> other and by setting the new flag in the service definition, clear intent
>> is communicated by the developers.
>>
>> What do you think about these two options?
>>
>> If you are already using this feature to create REST endpoints or plan to
>> use it in the future, please let me know as well, as it would mean that the
>> second option has merit and it is the least destructive change of the two.
>>
>> Thank you and best regards,
>> Konstantinos Marinos
>>
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Divesh Dutta <di...@hotwaxsystems.com>.
I agree with Mridul and Lukas.
For managing rest APIs, we are gradually moving away from service
definitions, so we can manage everything in rest.xml .
Last month Deepak also removed support for the "action" attribute from the
service definition in this commit:
https://github.com/apache/ofbiz-framework/commit/4fb03d40d1 and this:
https://github.com/apache/ofbiz-site/commit/bbfe081
Therefore, adding an attribute to the service definition is not a good
idea.
Thanks
--
Divesh Dutta
www.hotwaxsystems.com
On Wed, Jul 29, 2026 at 12:48 PM Lukas Finster <lu...@ecomify.de>
wrote:
> +1
>
> Am 29.07.26 um 09:14 schrieb Mridul Pathak:
> > I prefer the first approach. It's better to have a single way to expose
> any
> > service as a REST endpoint through rest.xml. Those already using the
> > current feature will need to migrate anyway.
> >
> > Thanks
> > Mridul Pathak
> >
> > On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <
> contact@kmarinos.com>
> > wrote:
> >
> >> Hi all,
> >>
> >> I was hoping to start a discussion about the current interaction of
> setting
> >> export="true" in a service definition in regards to the recently added
> >> rest-api module in the framework. The same flag that existed before
> >> (export="true") now additionally exposes a service definition as a REST
> >> endpoint. This is one of the ways to create a REST endpoint (the other
> >> major one being a *.rest.xml definition file) but since the component is
> >> now part of the framework, this could have unintended consequences.
> >>
> >> My biggest concern is that developers might not immediately realise that
> >> this one flag is used for similar but distinct use cases and this might
> not
> >> be the desired behaviour for all new or previously exported services.
> >>
> >> In order to avoid implicitly exposing services with potentially
> unintended
> >> consequences, I am reaching out for your thoughts on the following
> >> alternative actions:
> >>
> >> * We remove the feature of defining REST endpoints in this matter
> >> completely. Previously existing usages of export="true" remain
> unaffected
> >> and REST endpoints can be defined by dedicated rest.xml files.
> >>
> >> * We create a separate flag in the service definition (e.g.
> >> export-api="true") that only controls the auto discovery and creation of
> >> these REST endpoints. The two export features are then decoupled from
> each
> >> other and by setting the new flag in the service definition, clear
> intent
> >> is communicated by the developers.
> >>
> >> What do you think about these two options?
> >>
> >> If you are already using this feature to create REST endpoints or plan
> to
> >> use it in the future, please let me know as well, as it would mean that
> the
> >> second option has merit and it is the least destructive change of the
> two.
> >>
> >> Thank you and best regards,
> >> Konstantinos Marinos
> >>
> --
> Lukas Finster
> Softwareentwickler & Berater
>
> ecomify GmbH, Stralsunder Straße 63, 33605 Bielefeld
> Fon: +49 521 448157-90 | Fax: +49 521 448157-99 | www.ecomify.de
> Court Registration: Amtsgericht Bielefeld, HRB 41683 | CEO: Martin Becker,
> Michael Brohl
>
>
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Lukas Finster <lu...@ecomify.de>.
+1
Am 29.07.26 um 09:14 schrieb Mridul Pathak:
> I prefer the first approach. It's better to have a single way to expose any
> service as a REST endpoint through rest.xml. Those already using the
> current feature will need to migrate anyway.
>
> Thanks
> Mridul Pathak
>
> On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <co...@kmarinos.com>
> wrote:
>
>> Hi all,
>>
>> I was hoping to start a discussion about the current interaction of setting
>> export="true" in a service definition in regards to the recently added
>> rest-api module in the framework. The same flag that existed before
>> (export="true") now additionally exposes a service definition as a REST
>> endpoint. This is one of the ways to create a REST endpoint (the other
>> major one being a *.rest.xml definition file) but since the component is
>> now part of the framework, this could have unintended consequences.
>>
>> My biggest concern is that developers might not immediately realise that
>> this one flag is used for similar but distinct use cases and this might not
>> be the desired behaviour for all new or previously exported services.
>>
>> In order to avoid implicitly exposing services with potentially unintended
>> consequences, I am reaching out for your thoughts on the following
>> alternative actions:
>>
>> * We remove the feature of defining REST endpoints in this matter
>> completely. Previously existing usages of export="true" remain unaffected
>> and REST endpoints can be defined by dedicated rest.xml files.
>>
>> * We create a separate flag in the service definition (e.g.
>> export-api="true") that only controls the auto discovery and creation of
>> these REST endpoints. The two export features are then decoupled from each
>> other and by setting the new flag in the service definition, clear intent
>> is communicated by the developers.
>>
>> What do you think about these two options?
>>
>> If you are already using this feature to create REST endpoints or plan to
>> use it in the future, please let me know as well, as it would mean that the
>> second option has merit and it is the least destructive change of the two.
>>
>> Thank you and best regards,
>> Konstantinos Marinos
>>
--
Lukas Finster
Softwareentwickler & Berater
ecomify GmbH, Stralsunder Straße 63, 33605 Bielefeld
Fon: +49 521 448157-90 | Fax: +49 521 448157-99 | www.ecomify.de
Court Registration: Amtsgericht Bielefeld, HRB 41683 | CEO: Martin Becker, Michael Brohl
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Mridul Pathak <mr...@apache.org>.
I prefer the first approach. It's better to have a single way to expose any
service as a REST endpoint through rest.xml. Those already using the
current feature will need to migrate anyway.
Thanks
Mridul Pathak
On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <co...@kmarinos.com>
wrote:
> Hi all,
>
> I was hoping to start a discussion about the current interaction of setting
> export="true" in a service definition in regards to the recently added
> rest-api module in the framework. The same flag that existed before
> (export="true") now additionally exposes a service definition as a REST
> endpoint. This is one of the ways to create a REST endpoint (the other
> major one being a *.rest.xml definition file) but since the component is
> now part of the framework, this could have unintended consequences.
>
> My biggest concern is that developers might not immediately realise that
> this one flag is used for similar but distinct use cases and this might not
> be the desired behaviour for all new or previously exported services.
>
> In order to avoid implicitly exposing services with potentially unintended
> consequences, I am reaching out for your thoughts on the following
> alternative actions:
>
> * We remove the feature of defining REST endpoints in this matter
> completely. Previously existing usages of export="true" remain unaffected
> and REST endpoints can be defined by dedicated rest.xml files.
>
> * We create a separate flag in the service definition (e.g.
> export-api="true") that only controls the auto discovery and creation of
> these REST endpoints. The two export features are then decoupled from each
> other and by setting the new flag in the service definition, clear intent
> is communicated by the developers.
>
> What do you think about these two options?
>
> If you are already using this feature to create REST endpoints or plan to
> use it in the future, please let me know as well, as it would mean that the
> second option has merit and it is the least destructive change of the two.
>
> Thank you and best regards,
> Konstantinos Marinos
>
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Anahita Goljahani <an...@gmail.com>.
Hi Konstantinos, hi all,
I agree with the concern that using export="true" could
unintentionally expose services.
Both approaches you propose are fine with me: either introducing a
dedicated export-api="true" attribute or removing the feature
altogether.
However, one point I'd like to add is that, regardless of the removal
of the 'action' attribute by Deepak, using export="true" has never
really allowed the creation of a REST API in the architectural sense.
With that approach, you expose individual service definitions, but you
can't model a single endpoint as a resource and use the different HTTP
methods (GET, POST, PUT, DELETE, etc.) to operate on that resource.
Simply changing the name of the attribute to 'export-api' doesn't
fundamentally change the fact that this mechanism is not sufficient
for properly designing RESTful, resource-oriented APIs.
Regards
Anahita
Il giorno mar 28 lug 2026 alle ore 21:59 Konstantinos Marinos
<co...@kmarinos.com> ha scritto:
>
> Hi all,
>
> I was hoping to start a discussion about the current interaction of setting
> export="true" in a service definition in regards to the recently added
> rest-api module in the framework. The same flag that existed before
> (export="true") now additionally exposes a service definition as a REST
> endpoint. This is one of the ways to create a REST endpoint (the other
> major one being a *.rest.xml definition file) but since the component is
> now part of the framework, this could have unintended consequences.
>
> My biggest concern is that developers might not immediately realise that
> this one flag is used for similar but distinct use cases and this might not
> be the desired behaviour for all new or previously exported services.
>
> In order to avoid implicitly exposing services with potentially unintended
> consequences, I am reaching out for your thoughts on the following
> alternative actions:
>
> * We remove the feature of defining REST endpoints in this matter
> completely. Previously existing usages of export="true" remain unaffected
> and REST endpoints can be defined by dedicated rest.xml files.
>
> * We create a separate flag in the service definition (e.g.
> export-api="true") that only controls the auto discovery and creation of
> these REST endpoints. The two export features are then decoupled from each
> other and by setting the new flag in the service definition, clear intent
> is communicated by the developers.
>
> What do you think about these two options?
>
> If you are already using this feature to create REST endpoints or plan to
> use it in the future, please let me know as well, as it would mean that the
> second option has merit and it is the least destructive change of the two.
>
> Thank you and best regards,
> Konstantinos Marinos
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Chandan Khandelwal <ch...@hotwaxsystems.com>.
+1 for the first approach. I think having rest.xml as the dedicated way to
define REST endpoints makes the intent clearer and avoids unintentionally
exposing services.
Kind Regards,
Chandan Khandelwal
On Wed, Jul 29, 2026 at 5:34 PM Ashish Vijaywargiya <
ashish.vijaywargiya@hotwaxsystems.com> wrote:
> 1st approach looks good to me.
>
> --
> Kind Regards,
> Ashish Vijaywargiya
> Vice President of Operations
> *HotWax Systems*
> *Enterprise open source experts*
> http://www.hotwaxsystems.com
>
>
>
> On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <contact@kmarinos.com
> >
> wrote:
>
> > Hi all,
> >
> > I was hoping to start a discussion about the current interaction of
> setting
> > export="true" in a service definition in regards to the recently added
> > rest-api module in the framework. The same flag that existed before
> > (export="true") now additionally exposes a service definition as a REST
> > endpoint. This is one of the ways to create a REST endpoint (the other
> > major one being a *.rest.xml definition file) but since the component is
> > now part of the framework, this could have unintended consequences.
> >
> > My biggest concern is that developers might not immediately realise that
> > this one flag is used for similar but distinct use cases and this might
> not
> > be the desired behaviour for all new or previously exported services.
> >
> > In order to avoid implicitly exposing services with potentially
> unintended
> > consequences, I am reaching out for your thoughts on the following
> > alternative actions:
> >
> > * We remove the feature of defining REST endpoints in this matter
> > completely. Previously existing usages of export="true" remain unaffected
> > and REST endpoints can be defined by dedicated rest.xml files.
> >
> > * We create a separate flag in the service definition (e.g.
> > export-api="true") that only controls the auto discovery and creation of
> > these REST endpoints. The two export features are then decoupled from
> each
> > other and by setting the new flag in the service definition, clear intent
> > is communicated by the developers.
> >
> > What do you think about these two options?
> >
> > If you are already using this feature to create REST endpoints or plan to
> > use it in the future, please let me know as well, as it would mean that
> the
> > second option has merit and it is the least destructive change of the
> two.
> >
> > Thank you and best regards,
> > Konstantinos Marinos
> >
>
Re: [DISCUSS] Regarding export="true" for REST services
Posted by Ashish Vijaywargiya <as...@hotwaxsystems.com>.
1st approach looks good to me.
--
Kind Regards,
Ashish Vijaywargiya
Vice President of Operations
*HotWax Systems*
*Enterprise open source experts*
http://www.hotwaxsystems.com
On Wed, Jul 29, 2026 at 1:29 AM Konstantinos Marinos <co...@kmarinos.com>
wrote:
> Hi all,
>
> I was hoping to start a discussion about the current interaction of setting
> export="true" in a service definition in regards to the recently added
> rest-api module in the framework. The same flag that existed before
> (export="true") now additionally exposes a service definition as a REST
> endpoint. This is one of the ways to create a REST endpoint (the other
> major one being a *.rest.xml definition file) but since the component is
> now part of the framework, this could have unintended consequences.
>
> My biggest concern is that developers might not immediately realise that
> this one flag is used for similar but distinct use cases and this might not
> be the desired behaviour for all new or previously exported services.
>
> In order to avoid implicitly exposing services with potentially unintended
> consequences, I am reaching out for your thoughts on the following
> alternative actions:
>
> * We remove the feature of defining REST endpoints in this matter
> completely. Previously existing usages of export="true" remain unaffected
> and REST endpoints can be defined by dedicated rest.xml files.
>
> * We create a separate flag in the service definition (e.g.
> export-api="true") that only controls the auto discovery and creation of
> these REST endpoints. The two export features are then decoupled from each
> other and by setting the new flag in the service definition, clear intent
> is communicated by the developers.
>
> What do you think about these two options?
>
> If you are already using this feature to create REST endpoints or plan to
> use it in the future, please let me know as well, as it would mean that the
> second option has merit and it is the least destructive change of the two.
>
> Thank you and best regards,
> Konstantinos Marinos
>