Continuous Integration

[nohal]

Originally Posted by rgleason

Gilletarom, did you see the Zboss post?

I guess we are using CI for TDD!

Just so you don't have to search too far, "Test Driven Development"
https://www.hanselman.com/blog/AppVe...yToBehold.aspx

Rick...
We do a lot of things, but test driven development certainly is not one of them...

Pavel

[rgleason]

Ok, Pavel. I guess ZBoss had it a bit wrong.

Would you mind terribly, looking at this description at the beginning. It would be good to get it right?

https://opencpn.org/wiki/dokuwiki/do...us_integration

I think Gilletarom was using it for Alpha testing!

What should we title this? Alpha Testing Tool?

[nohal]

Originally Posted by rgleason

Ok, Pavel. I guess ZBoss had it a bit wrong.

Would you mind terribly, looking at this description at the beginning. It would be good to get it right?

https://opencpn.org/wiki/dokuwiki/do...us_integration

I think Gilletarom was using it for Alpha testing!

What should we title this?

Rick...
I honestly do not understand the purpose of that page. What are we're trying to tell people there?
The main purpose of having these builds for us is to simply see whether we didn't break the build on some of the platforms.
For user testing, we release and announce builds for alpha and beta versions when it makes sense.
There may be situations when testing the builds coming from Appveyor makes sense, but do not make it look like something we want "normal public" to do. We don't. When it does make sense, we ask for it and point you to the build(s) suitable for it.

Pavel

[rgleason]

Ok, Pavel. You want me to delete the page, so you can describe how each time?
This is in the Development Manual.

[nohal]

Originally Posted by rgleason

Ok, Pavel. You want me to delete the page, so you can describe how each time?
This is in the Development Manual.

Rick, I just want to understand what we are trying to say by that page, that's all. And yes, I would much prefer if you discussed whether it makes sense to add stuff to the developer manual before actually adding it as I really don't think accumulating heaps of text that we are not able to maintain makes sense.

Pavel

[rgleason]

I really don't think accumulating heaps of text that we are not able to maintain makes sense.

I agree with you on this.

So you expect Programmers to know this stuff and they are the only ones who should know it? There are other people involved too. We're always looking for volunteers to test. There aren't enough of us as it is.

So I guess the real question is who is the development pages for?
1. Programmers know everything anyway, so they don't need it.
2. Public doesn't need it because they aren't interested. The program runs or not.
3. Who else is there to consider? I am at a loss.

[nohal]

Originally Posted by rgleason

I agree with you on this.

So you expect Programmers to know this stuff and they are the only ones who should know it? There are other people involved too. We're always looking for volunteers to test. There aren't enough of us as it is.

So I guess the real question is who is the development pages for?
1. Programmers know everything anyway, so they don't need it.
2. Public doesn't need it because they aren't interested. The program runs or not.
3. Who else is there to consider? I am at a loss.

Rick...
I expect a book called "Developer Manual" to be a manual and reference for developers. As you very correctly say, there aren't enough of us as it is, we simply can not provide all the information to take people from zero to being a programmer. There are other platforms that do it, like the educational system most of the countries have or the endless number of online educational sites.

Pavel

[rgleason]

I assume this is a discussion, because this is a forum. Pavel you and others have been building a software engine which confirms that code does compile. That is very helpful to a programmer because it confirms that his writing skills are "accepted" however it does not confirm that the program actually works.

For that purpose (at least for OpenCPN) you are relying on humans, Testers. The pool of Testers has been somewhat limited to those who can compile in some development environment (highly restrictedt). This new Testing Tool broadens the available number of testers. Why wouldn't you want to make it easier for them to help? - This is exactly Gillatarom's point.

I submit that the "Development Manual" is as much for a group of "advanced user" as for Developers. Perhaps we should rename it? Since Developers don't need it!

Perhaps it should be renamed Beta Manual, since developers don't need it.
At this point I am ready to "Go outside and play."

PS: I had thought that the term "Developer" had a much broader tent. I guess it is a term exclusively reserved for those who have been "educated" in the normal manner. Otherwise I never would have named it "Developer Manual" when we rebuilt the wiki.

[nohal]

Originally Posted by rgleason

I assume this is a discussion, because this is a forum. Pavel you and others have been building a software engine which confirms that code does compile. That is very helpful to a programmer because it confirms that his writing skills are "accepted" however it does not confirm that the program actually works.

For that purpose (at least for OpenCPN) you are relying on humans, Testers. The pool of Testers has been somewhat limited to those who can compile in some development environment (highly restrictedt). This new Testing Tool broadens the available number of testers. Why wouldn't you want to make it easier for them to help? - This is exactly Gillatarom's point.
Th
I submit that the "Development Manual" is as much for a group of "advanced user" as for Developers. Perhaps we should rename it?
Since Developers don't need it!

Developers of course do need it. Except the ones that actually do write it. But we need information in it being correct and accurate. That is hard to achieve if it keeps to be bloated with stuff like this page or several copies of build instruction filled with stuff you copy from the forum without actually understanding what it talks about.
I know your intentions are the purest, but I really don't feel comfortable about the results in this case.
Please, do what you can do and do well - test and document the actual functionality of the program, but leave the developer documentation to be written by people who actually know what they write about. And if the result is not clear or exhaustive enough, ask for clarifications, of course.

Pavel

[bdbcat]

Rick...

Let's not get too stirred up about CI.

We have lots of human testers, of widely different skill sets and experience. Some (e.g. you, Gilletarom, etc.) are experienced with OCPN, are capable, and proven adventurous. Others, not so much.

Back in the day, we called these early testers "heat-seekers".

For this first class of testers, using the CI products for informal alpha testing of unreleased code is very useful to the devs. The main benefit to devs is the rapid cross-platform exercise of the app, where testers bang away on the interface using their favorite workflows, and attempt to break it. Most devs don't keep current images of all 5 platforms available for extended testing, and in any case this takes a lot of time.

It simply must be understood that this alpha testing is ad-hoc, and is not tracked in any persistent manner. But it does advance the code-base integrity quickly.

So maybe we need a wiki topic called "Advanced testing tips", or some such?

and btw, this is a discussion.

Dave

[rgleason]

Thanks Pavel and Dave, I'm going to let this sit a bit if you don't mind & come back to it later.

Gilletarom, I am very interested in Weather_routing, and am intrigued that you used did-g's repos with the dev branch and the appveyor build to test his versions of weather_routing. How novel! I'd like to try it too. Were you using version 4.99 as first presented or some other version? I'd like to do a parallel install of the right version of Opencpn, so I can try did-g's version of weather_routing_pi. Thanks.

PS Since I am keeping my development environment at v4.8.2 with wxwidgets 3.0.2 and msvc++2013, as Pavel suggested earlier. So I am unable to compile these newer versions of Opencpn, so that's a situation where the Appveyor engine can really help!

[Gilletarom]

Hello,

Originally Posted by rgleason

....
Gilletarom, I am very interested in Weather_routing, and am intrigued that you used did-g's repos with the dev branch and the appveyor build to test his versions of weather_routing. How novel! I'd like to try it too. Were you using version 4.99 as first presented or some other version? I'd like to do a parallel install of the right version of Opencpn, so I can try did-g's version of weather_routing_pi. Thanks.
....

I do not have the power to give appVeyor access to did-g. Only Did-g can do it, if he wishes.

This refers to my questioning in my previous post, on this thread, about the merits of a free access or not to integrated development tools such as "Appveyor" or "Trevis CI". In my opinion,
- the developer, powner of the storage at "Appveyor" or "Trevis CI", in this case Did-g,
- the plugin manager, in this case Sean D'Epagnier for WR and Jongouh for DR,
- and perhaps Dave
are in the best position to decide.

It seems to me that the organization of the test method, and the choice of testers, can not be decided by a simple beta tester like me.

[rgleason]

Gilletarom
So you prefer unopen OpenCPN, and I do not agree with you or Pavel.

You decided not to answer my question, what version of OpenCPN did you use?
but never mind I will find out.

Such a friendly, cooperative, respectful and open environment.

[Gilletarom]

Originally Posted by rgleason

Gilletarom
So you prefer unopen OpenCPN, and I do not agree with you or Pavel.

You decided not to answer my question, what version of OpenCPN did you use?
but never mind I will find out.

Such a friendly, cooperative, respectful and open environment.

I think you misunderstand my intentions.

1
On the one hand, I had to give Did-g information from which Did-g opened for me access to his "AppVeyor". This is personal information and if tomorrow Did-g opens the same access door, you will have to give him your personal information.
2 ° Did-g manufactured, during this test period, spread over a month and a half, several versions of OpenCPN successive from the source of OpenCPN. Each version took into account, if I understand correctly, changes in the source code of OpenCPN. The latest is probably 4.99 because it displays the MBtiles.

In my intellectual functioning, from the moment Did-g trusts me, I must respect that confidence. Did-g does not have to publish its versions, I do not have to publish them.
I add, reading your reaction, that I am even beginning to reproach myself for having spoken about it.

[rgleason]

Gilletarom, such are words. Perhaps I misunderstood you.
No you can't violate did-g confidence. But we don't want to start trade wars when we are collaborating! Trade wars are a waste of energy and time. I have this file from Appveyor for did-g weather_routing (his dev branch)

https://ci.appveyor.com/project/did-...0.59/artifacts

I have downloaded it and installed it into 4.99 and 3.8.2
From the appveyor log I see that it is compiled with opencpn.lib from v4.8.2
When I run 4.99 the pi does not load.
When I run 4.8.2 the pi does load, but the PI will not compile with gribs and ocpn_draw boundaries do not seem to work, it does work with climatology. I have looked at the menus and can not determine if there are any changes.

I am interested in did-g's version because I was Sean's first guinea pig at Sean's inception and it is one of the things I wanted OpenCPN to do. Back then I had to compile it using mingw and a special version of OpenCPN because that was what Sean was using for Windows.