Stable trunk?

[Ritesh Trivedi]

Hi,

Let me first acknowledge that I made a mistake from the beginning to go with the latest from the trunk and not 4.0 which was the stable release. I have not tried 4.0 again to see the differences between 4.0 and the latest. Besides I have a few relatively minor changes that I have made into my local ofbiz copy.

Now the problem... I have been updating the latest since then to get the fixes for the things broken due to new checkins and i am going in circles, new patches seem to break a few other things and so on.

Question is - How carefully are checkins being made/accepted? (just out of curiosity) and does anyone know of a version - still close to latest that is relatively stable?

Thanks

[Adam Heath-2]

Ritesh Trivedi wrote:

> Hi,
>
> Let me first acknowledge that I made a mistake from the beginning to go with
> the latest from the trunk and not 4.0 which was the stable release. I have
> not tried 4.0 again to see the differences between 4.0 and the latest.
> Besides I have a few relatively minor changes that I have made into my local
> ofbiz copy.
>
> Now the problem... I have been updating the latest since then to get the
> fixes for the things broken due to new checkins and i am going in circles,
> new patches seem to break a few other things and so on.
>
> Question is - How carefully are checkins being made/accepted? (just out of
> curiosity) and does anyone know of a version - still close to latest that is
> relatively stable?

As is always the case, please report problems you have; there is no way
we can debug what issues you are having unless you tell us.

[BJ Freeman]

Short answer trunk is not stable. That is when you are only doing bug
fixes not adding new features. Some new Features are being added in
sections.
1) can't mix and match 4.0 and trunk. 4.0 uses 1.42 trunk uses 1.5+
2) You can go over the commit ML and get each commit to see what is
changed. Yes the committers are suppose to check each change. However
with the size of ofbiz, that usually is focused, on the area the patch
effects, not all of ofbiz.
Some committers seem to add code then do testing.
when you find a bug test on the demo server, if you can replicate it
there put in a jira.
http://www.apache.org/dev/committers.html

Ritesh Trivedi sent the following on 10/27/2008 7:25 PM:

> Hi,
>
> Let me first acknowledge that I made a mistake from the beginning to go with
> the latest from the trunk and not 4.0 which was the stable release. I have
> not tried 4.0 again to see the differences between 4.0 and the latest.
> Besides I have a few relatively minor changes that I have made into my local
> ofbiz copy.
>
> Now the problem... I have been updating the latest since then to get the
> fixes for the things broken due to new checkins and i am going in circles,
> new patches seem to break a few other things and so on.
>
> Question is - How carefully are checkins being made/accepted? (just out of
> curiosity) and does anyone know of a version - still close to latest that is
> relatively stable?
>
> Thanks

[Jacques Le Roux]

From: "BJ Freeman" <[hidden email]>
> Some committers seem to add code then do testing.
> when you find a bug test on the demo server, if you can replicate it
> there put in a jira.

This is a very unsual case. We call it CTR (Commit then Review) mode and it's not much used in OFBiz. Sometimes though some bugs
slip in, we are humans. This is why reporting bugs is so important. With a patch is much appreciated ;o)

Jacques

> http://www.apache.org/dev/committers.html
>
> Ritesh Trivedi sent the following on 10/27/2008 7:25 PM:
>> Hi,
>>
>> Let me first acknowledge that I made a mistake from the beginning to go with
>> the latest from the trunk and not 4.0 which was the stable release. I have
>> not tried 4.0 again to see the differences between 4.0 and the latest.
>> Besides I have a few relatively minor changes that I have made into my local
>> ofbiz copy.
>>
>> Now the problem... I have been updating the latest since then to get the
>> fixes for the things broken due to new checkins and i am going in circles,
>> new patches seem to break a few other things and so on.
>>
>> Question is - How carefully are checkins being made/accepted? (just out of
>> curiosity) and does anyone know of a version - still close to latest that is
>> relatively stable?
>>
>> Thanks
>

[Ritesh Trivedi]

Is there a minor release on horizon? or is there a release map somewhere which shows roadmap, projected milestones, dates etc?

jacques.le.roux wrote

From: "BJ Freeman" <[e-mail]>
> Some committers seem to add code then do testing.
> when you find a bug test on the demo server, if you can replicate it
> there put in a jira.

This is a very unsual case. We call it CTR (Commit then Review) mode and it's not much used in OFBiz. Sometimes though some bugs
slip in, we are humans. This is why reporting bugs is so important. With a patch is much appreciated ;o)

Jacques

> http://www.apache.org/dev/committers.html
>
> Ritesh Trivedi sent the following on 10/27/2008 7:25 PM:
>> Hi,
>>
>> Let me first acknowledge that I made a mistake from the beginning to go with
>> the latest from the trunk and not 4.0 which was the stable release. I have
>> not tried 4.0 again to see the differences between 4.0 and the latest.
>> Besides I have a few relatively minor changes that I have made into my local
>> ofbiz copy.
>>
>> Now the problem... I have been updating the latest since then to get the
>> fixes for the things broken due to new checkins and i am going in circles,
>> new patches seem to break a few other things and so on.
>>
>> Question is - How carefully are checkins being made/accepted? (just out of
>> curiosity) and does anyone know of a version - still close to latest that is
>> relatively stable?
>>
>> Thanks
>

[Vince Clark]

There is a process called "Vendor Branch Management" that you can read about in the Subversion book. If you want to maintain your own stable version and occasionally merge with revisions from trunk you can use this process.

It is a pretty good process in my opinion but also very resource intensive. And you would still need to do your own QA and functional testing in order to determine what revisions are worthy of merging.

----- Original Message -----
From: "Ritesh Trivedi" <[hidden email]>
To: [hidden email]
Sent: Tuesday, October 28, 2008 10:40:51 AM (GMT-0700) America/Denver
Subject: Re: Stable trunk?


Is there a minor release on horizon? or is there a release map somewhere
which shows roadmap, projected milestones, dates etc?


jacques.le.roux wrote:

>
> From: "BJ Freeman" <[hidden email]>
>> Some committers seem to add code then do testing.
>> when you find a bug test on the demo server, if you can replicate it
>> there put in a jira.
>
> This is a very unsual case. We call it CTR (Commit then Review) mode and
> it's not much used in OFBiz. Sometimes though some bugs
> slip in, we are humans. This is why reporting bugs is so important. With a
> patch is much appreciated ;o)
>
> Jacques
>
>> http://www.apache.org/dev/committers.html 
>>
>> Ritesh Trivedi sent the following on 10/27/2008 7:25 PM:
>>> Hi,
>>>
>>> Let me first acknowledge that I made a mistake from the beginning to go
>>> with
>>> the latest from the trunk and not 4.0 which was the stable release. I
>>> have
>>> not tried 4.0 again to see the differences between 4.0 and the latest.
>>> Besides I have a few relatively minor changes that I have made into my
>>> local
>>> ofbiz copy.
>>>
>>> Now the problem... I have been updating the latest since then to get the
>>> fixes for the things broken due to new checkins and i am going in
>>> circles,
>>> new patches seem to break a few other things and so on.
>>>
>>> Question is - How carefully are checkins being made/accepted? (just out
>>> of
>>> curiosity) and does anyone know of a version - still close to latest
>>> that is
>>> relatively stable?
>>>
>>> Thanks
>>
>
>
>

--
View this message in context: http://www.nabble.com/Stable-trunk--tp20200392p20211017.html 
Sent from the OFBiz - Dev mailing list archive at Nabble.com.

[BJ Freeman]

I see this in the commit logs continuously.
I would think if someone did thorough testing before commit that this
would not happen.
No biggy to me, since i have not brought it up till now.
just a way to explain bugs and to check the commits to see if it is
fixed before submitting a jira.


Jacques Le Roux sent the following on 10/28/2008 9:26 AM:

> From: "BJ Freeman" <[hidden email]>
>> Some committers seem to add code then do testing.
>> when you find a bug test on the demo server, if you can replicate it
>> there put in a jira.
>
> This is a very unsual case. We call it CTR (Commit then Review) mode and
> it's not much used in OFBiz. Sometimes though some bugs slip in, we are
> humans. This is why reporting bugs is so important. With a patch is much
> appreciated ;o)
>
> Jacques
>
>> http://www.apache.org/dev/committers.html
>>
>> Ritesh Trivedi sent the following on 10/27/2008 7:25 PM:
>>> Hi,
>>>
>>> Let me first acknowledge that I made a mistake from the beginning to
>>> go with
>>> the latest from the trunk and not 4.0 which was the stable release. I
>>> have
>>> not tried 4.0 again to see the differences between 4.0 and the latest.
>>> Besides I have a few relatively minor changes that I have made into
>>> my local
>>> ofbiz copy.
>>>
>>> Now the problem... I have been updating the latest since then to get the
>>> fixes for the things broken due to new checkins and i am going in
>>> circles,
>>> new patches seem to break a few other things and so on.
>>>
>>> Question is - How carefully are checkins being made/accepted? (just
>>> out of
>>> curiosity) and does anyone know of a version - still close to latest
>>> that is
>>> relatively stable?
>>>
>>> Thanks
>>
>
>
>

[Jeroen van der Wal]

Couldn't we use the principles of Test Driven Development as an approach to get and keep the trunk "stable"? I plea for more and better use of unit tests and not to commit code that does not pass the tests. It also makes it easier for reviewers to do what they have to do: just review and not test.

Introducing TDD in my company greatly reduced the amount of bugs and made use create new releases with very little effort. I understand that controlled environment is different from a community driven project so I'm curious what your opinions are.

-Jeroen

BJ Freeman wrote

Short answer trunk is not stable. That is when you are only doing bug
fixes not adding new features. Some new Features are being added in
sections.
1) can't mix and match 4.0 and trunk. 4.0 uses 1.42 trunk uses 1.5+
2) You can go over the commit ML and get each commit to see what is
changed. Yes the committers are suppose to check each change. However
with the size of ofbiz, that usually is focused, on the area the patch
effects, not all of ofbiz.
Some committers seem to add code then do testing.
when you find a bug test on the demo server, if you can replicate it
there put in a jira.
http://www.apache.org/dev/committers.html

Ritesh Trivedi sent the following on 10/27/2008 7:25 PM:
> Hi,
>
> Let me first acknowledge that I made a mistake from the beginning to go with
> the latest from the trunk and not 4.0 which was the stable release. I have
> not tried 4.0 again to see the differences between 4.0 and the latest.
> Besides I have a few relatively minor changes that I have made into my local
> ofbiz copy.
>
> Now the problem... I have been updating the latest since then to get the
> fixes for the things broken due to new checkins and i am going in circles,
> new patches seem to break a few other things and so on.
>
> Question is - How carefully are checkins being made/accepted? (just out of
> curiosity) and does anyone know of a version - still close to latest that is
> relatively stable?
>
> Thanks

[BJ Freeman]

there has been an effort to put in test units.
the only thing lacking, in making it complete is manpower.

Jeroen van der Wal sent the following on 10/29/2008 8:47 AM:

> Couldn't we use the principles of Test Driven Development as an approach to
> get and keep the trunk "stable"? I plea for more and better use of unit
> tests and not to commit code that does not pass the tests. It also makes it
> easier for reviewers to do what they have to do: just review and not test.
>
> Introducing TDD in my company greatly reduced the amount of bugs and made
> use create new releases with very little effort. I understand that
> controlled environment is different from a community driven project so I'm
> curious what your opinions are.
>
> -Jeroen
>
>
> BJ Freeman wrote:
>> Short answer trunk is not stable. That is when you are only doing bug
>> fixes not adding new features. Some new Features are being added in
>> sections.
>> 1) can't mix and match 4.0 and trunk. 4.0 uses 1.42 trunk uses 1.5+
>> 2) You can go over the commit ML and get each commit to see what is
>> changed. Yes the committers are suppose to check each change. However
>> with the size of ofbiz, that usually is focused, on the area the patch
>> effects, not all of ofbiz.
>> Some committers seem to add code then do testing.
>> when you find a bug test on the demo server, if you can replicate it
>> there put in a jira.
>> http://www.apache.org/dev/committers.html
>>
>> Ritesh Trivedi sent the following on 10/27/2008 7:25 PM:
>>> Hi,
>>>
>>> Let me first acknowledge that I made a mistake from the beginning to go
>>> with
>>> the latest from the trunk and not 4.0 which was the stable release. I
>>> have
>>> not tried 4.0 again to see the differences between 4.0 and the latest.
>>> Besides I have a few relatively minor changes that I have made into my
>>> local
>>> ofbiz copy.
>>>
>>> Now the problem... I have been updating the latest since then to get the
>>> fixes for the things broken due to new checkins and i am going in
>>> circles,
>>> new patches seem to break a few other things and so on.
>>>
>>> Question is - How carefully are checkins being made/accepted? (just out
>>> of
>>> curiosity) and does anyone know of a version - still close to latest that
>>> is
>>> relatively stable?
>>>
>>> Thanks
>>
>

[Adam Heath-2]

BJ Freeman wrote:
> there has been an effort to put in test units.
> the only thing lacking, in making it complete is manpower.

And fixing the existing tests that are broken. :|

[David E Jones]

On Oct 29, 2008, at 11:54 AM, Adam Heath wrote:

> BJ Freeman wrote:
>> there has been an effort to put in test units.
>> the only thing lacking, in making it complete is manpower.
>
> And fixing the existing tests that are broken. :|

This is an area where it would be REALLY GREAT to have more effort go  
into the project. Yep, great enough to capitalize "REALLY" and "GREAT".

Who has worked on the unit tests that are in place? I'll admit I  
haven't much except on the toolset and some of the framework unit  
tests and helping some of the Hotwax Media people who wrote many of  
the tests that now exist, especially the ones in the various  
applications.

Is there anyone interested in working on this stuff? If there are  
enough people who want to actively work on it we can setup some  
coordination resources (ie Jira tasks, confluence pages, etc). If  
there are only 2-3 then coordination through the mailing list would be  
better, and more visible to others possibly interested.

-David

[Erwan de FERRIERES-3]

Here at Nereide, we are ready to make selenium tests (it's a task we
have planned to do, but which is always postponed....).
So, if it's ok with you and that you are interested in that, we are
going to make it real !

David E Jones a écrit :

>
> On Oct 29, 2008, at 11:54 AM, Adam Heath wrote:
>
>> BJ Freeman wrote:
>>> there has been an effort to put in test units.
>>> the only thing lacking, in making it complete is manpower.
>>
>> And fixing the existing tests that are broken. :|
>
> This is an area where it would be REALLY GREAT to have more effort go
> into the project. Yep, great enough to capitalize "REALLY" and "GREAT".
>
> Who has worked on the unit tests that are in place? I'll admit I haven't
> much except on the toolset and some of the framework unit tests and
> helping some of the Hotwax Media people who wrote many of the tests that
> now exist, especially the ones in the various applications.
>
> Is there anyone interested in working on this stuff? If there are enough
> people who want to actively work on it we can setup some coordination
> resources (ie Jira tasks, confluence pages, etc). If there are only 2-3
> then coordination through the mailing list would be better, and more
> visible to others possibly interested.
>
> -David
>
>

--
- Erwan -

[BJ Freeman]

I have been slowly creating test more from a user input than testing code.
I think I can complete one that will work on the trunk for an example
and review.
target jan 09

David E Jones sent the following on 10/29/2008 11:02 AM:

>
> On Oct 29, 2008, at 11:54 AM, Adam Heath wrote:
>
>> BJ Freeman wrote:
>>> there has been an effort to put in test units.
>>> the only thing lacking, in making it complete is manpower.
>>
>> And fixing the existing tests that are broken. :|
>
> This is an area where it would be REALLY GREAT to have more effort go
> into the project. Yep, great enough to capitalize "REALLY" and "GREAT".
>
> Who has worked on the unit tests that are in place? I'll admit I haven't
> much except on the toolset and some of the framework unit tests and
> helping some of the Hotwax Media people who wrote many of the tests that
> now exist, especially the ones in the various applications.
>
> Is there anyone interested in working on this stuff? If there are enough
> people who want to actively work on it we can setup some coordination
> resources (ie Jira tasks, confluence pages, etc). If there are only 2-3
> then coordination through the mailing list would be better, and more
> visible to others possibly interested.
>
> -David
>
>
>

[Shi Yusen]

I have some spare time in next 3 months and can code some test cases.

Shi Yusen/Beijing Langhua Ltd.


在 2008-10-29三的 12:02 -0600，David E Jones写道：

> On Oct 29, 2008, at 11:54 AM, Adam Heath wrote:
>
> > BJ Freeman wrote:
> >> there has been an effort to put in test units.
> >> the only thing lacking, in making it complete is manpower.
> >
> > And fixing the existing tests that are broken. :|
>
> This is an area where it would be REALLY GREAT to have more effort go  
> into the project. Yep, great enough to capitalize "REALLY" and "GREAT".
>
> Who has worked on the unit tests that are in place? I'll admit I  
> haven't much except on the toolset and some of the framework unit  
> tests and helping some of the Hotwax Media people who wrote many of  
> the tests that now exist, especially the ones in the various  
> applications.
>
> Is there anyone interested in working on this stuff? If there are  
> enough people who want to actively work on it we can setup some  
> coordination resources (ie Jira tasks, confluence pages, etc). If  
> there are only 2-3 then coordination through the mailing list would be  
> better, and more visible to others possibly interested.
>
> -David
>

[Adam Heath-2]

Shi Yusen wrote:
> I have some spare time in next 3 months and can code some test cases.

David was talking about fixing the *existing* tests, making them pass,
or fixing the code that has broken them.

[Shi Yusen]

Thanks Adam!

Surprising and interesting.:)

What's the latest trunk version that I can build successfully? I'll try
to do some test run later.


在 2008-10-29三的 13:56 -0500，Adam Heath写道：
> Shi Yusen wrote:
> > I have some spare time in next 3 months and can code some test cases.
>
> David was talking about fixing the *existing* tests, making them pass,
> or fixing the code that has broken them.

[Jeroen van der Wal]

We're currently working on unit testing for our custom (OFBiz) application and I'm willing to work on this.

-Jeroen

David E Jones wrote

On Oct 29, 2008, at 11:54 AM, Adam Heath wrote:

> BJ Freeman wrote:
>> there has been an effort to put in test units.
>> the only thing lacking, in making it complete is manpower.
>
> And fixing the existing tests that are broken. :|

This is an area where it would be REALLY GREAT to have more effort go  
into the project. Yep, great enough to capitalize "REALLY" and "GREAT".

Who has worked on the unit tests that are in place? I'll admit I  
haven't much except on the toolset and some of the framework unit  
tests and helping some of the Hotwax Media people who wrote many of  
the tests that now exist, especially the ones in the various  
applications.

Is there anyone interested in working on this stuff? If there are  
enough people who want to actively work on it we can setup some  
coordination resources (ie Jira tasks, confluence pages, etc). If  
there are only 2-3 then coordination through the mailing list would be  
better, and more visible to others possibly interested.

-David

[Scott Gray]

From memory, the main reason the current unit tests are failing is
because they were mistakenly only tested against the component they
were built for and not when running the entire set of tests.  It means
that the demo data used in the tests gets reused across tests and as
that data changes it causes subsequent tests to fail because they were
expected the data to be in it's original state.

In most cases simply adding more test data and directing the later
tests to use that instead should solve most of the problems.

Regards
Scott

2008/10/30 Shi Yusen <[hidden email]>:

> Thanks Adam!
>
> Surprising and interesting.:)
>
> What's the latest trunk version that I can build successfully? I'll try
> to do some test run later.
>
>
> 在 2008-10-29三的 13:56 -0500，Adam Heath写道：
>> Shi Yusen wrote:
>> > I have some spare time in next 3 months and can code some test cases.
>>
>> David was talking about fixing the *existing* tests, making them pass,
>> or fixing the code that has broken them.
>
>

[David E Jones]

Erwan,

It would be great to have some tests that go through the user  
interface, but we don't have tools for this yet that fit into the  
automated test system in OFBiz (ie so they can run along with other  
tests, and run automatically).

The goal is for the tests to all work with an "ant run-tests" (or  
"java -jar ofbiz.jar tests"), and to cover as much of OFBiz OOTB as  
possible, and then to also be easy to customize or comment out those  
that no longer apply after people customize or add on to OFBiz.

If you guys would like to work on getting Selenium tests to work this  
way, that would be great. Others have looked at this and run into  
troubles, so the last idea I heard was to use something different and  
that might be more manual for initial test writing, but probably  
easier to maintain.

At Hotwax we've written/recorded a bunch of Selenium tests for  
clients, but they are difficult to maintain and as far as we've gone  
they also have to be manually run and watched.

-David


On Oct 29, 2008, at 12:09 PM, Erwan de FERRIERES wrote:

> Here at Nereide, we are ready to make selenium tests (it's a task we  
> have planned to do, but which is always postponed....).
> So, if it's ok with you and that you are interested in that, we are  
> going to make it real !
>
> David E Jones a écrit :
>> On Oct 29, 2008, at 11:54 AM, Adam Heath wrote:
>>> BJ Freeman wrote:
>>>> there has been an effort to put in test units.
>>>> the only thing lacking, in making it complete is manpower.
>>>
>>> And fixing the existing tests that are broken. :|
>> This is an area where it would be REALLY GREAT to have more effort  
>> go into the project. Yep, great enough to capitalize "REALLY" and  
>> "GREAT".
>> Who has worked on the unit tests that are in place? I'll admit I  
>> haven't much except on the toolset and some of the framework unit  
>> tests and helping some of the Hotwax Media people who wrote many of  
>> the tests that now exist, especially the ones in the various  
>> applications.
>> Is there anyone interested in working on this stuff? If there are  
>> enough people who want to actively work on it we can setup some  
>> coordination resources (ie Jira tasks, confluence pages, etc). If  
>> there are only 2-3 then coordination through the mailing list would  
>> be better, and more visible to others possibly interested.
>> -David
>
> --
> - Erwan -

[David E Jones]

On Oct 29, 2008, at 12:56 PM, Adam Heath wrote:

> Shi Yusen wrote:
>> I have some spare time in next 3 months and can code some test cases.
>
> David was talking about fixing the *existing* tests, making them pass,
> or fixing the code that has broken them.

Yes, this would be the first priority.

Other priorities would include adding more unit tests for the  
framework and applications, and enhancing the test infrastructure to  
support automated UI tests (which I just mentioned in another reply).

There are LOTS of framework features that still don't have unit tests,  
though those don't tend to change as much so I'd still say more  
complete application unit tests are a higher priority. BTW, when I say  
application unit tests I really mean testing services and other logic  
level stuff in OFBiz. This is easiest to do by walking through  
processes in applications and seeing which services are called at each  
step, and then calling those services in the tests. More exotic ones  
could explore outside of that and do more tests on individual  
services, or tests processes that are not yet supported in the UI.

Of course, this is a volunteer-driven effort, and hopefully the  
"volunteer" efforts will be driven by actual needs different  
organizations have that are generic enough to be contributed back  
(those tend to be the most useful contributions, ie very "grounded" in  
real world needs).

-David