Continuous Integration

[rgleason]

Continuous Integration with Travis CI and Appeveyor for Windows looks like it is really going to help with testing becase it makes newer code availale without having to compile it yourself.

1. How to get Beta Testers familiar with Github and locating those compiled files?
2. Are these eventually going to be the source for the Website Downloads page?

Thanks Pavel and others for building this nice software engine!

PS: I notice that Chocolate on sourceforge often fails.

[bdbcat]

Rick...

1. How to get Beta Testers familiar with Github and locating those compiled files?

When we are actually in Beta test, we will describe the process and condition of the CI artifacts. We may use the artifacts then, but probably not. We have enough trouble tracking issues with one "official" Beta location. Using the almost daily updates of CI product for Beta testing would be a nightmare.

Best to think of the usefulness of CI as we are right now: In Alpha test of some new features, some of which features are known to be incomplete or even with known heinous errors.

2. Are these eventually going to be the source for the Website Downloads page?

No. The CI artifacts result from building the current github HEAD. This, in general, will not be the same as the in-production Release point.

So, to summarize:
Testing a CI Product is like sneaking a spoon-sip of the dinner simmering on the stove while Mom is not looking. She knows it needs salt, and is way too hot to eat yet. If you drop dead immediately, she wants to know about it. Otherwise, she wants to not hear too many comments. Wait until it comes to the table. Meanwhile, go outside and play....


Thanks
Dave

[sailorF54]

Originally Posted by bdbcat

Testing a CI Product is like...
....Meanwhile, go outside and play....

[rgleason]

Dave. Thanks for the analogy and explanation.

[rgleason]

I've added a page under Dev, may add a screenshot with your permission.

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

[bdbcat]

Rick...

Screenshot is fine.

Thanks for the doku update!

Dave

[rgleason]

Here is a post about CI . I'll include something along these lines.

http://www.cruisersforum.com/forums/...ml#post2577904

[rgleason]

Hope this is clear enough.
https://opencpn.org/wiki/dokuwiki/do...us_integration

[Gilletarom]

Hello RGleason, hello Dave, hello All,

Hum ... CI ? CI ? CI ? It took me a day, and the two last post of RGleason, to understand that CI meant in your head "Continuous integration".

Hum .... You live in your world.

And so, of course, in the meantime, OCPN is free software, ie whose code is public.

But in reality, the complexity of the operation of development becomes "monstrous"

And because of this, everything happens as if OpenCPN was a "close source" product.

Is this the price to pay, so that OpenCPN remains OpenCPN?

[rgleason]

Sorry Gilletarom,

one of them is "Travis CI" I guess I should not have shortened it.

[bdbcat]

Gilletarom...

Is this the price to pay, so that OpenCPN remains OpenCPN?

You make good points.

The day is long gone where a meaningful application consists of a few thousand lines of code, readable at one sitting. Now, one should (must?) take advantage of all possible advances in the programmer's arsenal of tools, lest the app and its environment become quickly stale.

Added complexity is a pity, really. Believe me, most devs I know often long for the good old days of single-platform coding. Seems like we spend more time fussing with tools and obscure APIs than we do writing base code algorithms. Sigh....

But this is really a price to pay for progress, and we plan for OCPN to stay as current and timely as we are able to make it.

Thanks for your comments.
Dave

[seandepagnier]

Originally Posted by Gilletarom

Hello RGleason, hello Dave, hello All,
And because of this, everything happens as if OpenCPN was a "close source" product.

Maybe it takes a long time, but at least you have the right to read the source, and if you do not have the time, others do, and so it is nothing like "close source"

[Gilletarom]

Hello Sean D'Epagnier,

Originally Posted by boat_alexandra

Maybe it takes a long time, but at least you have the right to read the source, and if you do not have the time, others do, and so it is nothing like "close source"

Sorry. In truth, you are right. That said, I ran into my post my ignorance of the English language (or American). And by writing {as if OpenCPN was a "close source" product}, I have only tried to say that: "OpenCPN was becoming a product increasingly difficult to access".

[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"

See
AppVeyor - A good continuous integration system is a joy to behold

You are getting continuous integration and test driven development mixed up. TDD is a common aspect of CI but lots of devshops does one or the other. Don't even bring in devtestops.

What is most commonly misunderstood about CI and devops is that they are some how the wild west when in fact they are highly coordinated activities. The most effective CI/CD system compress the release trains down into a single push and everyone is developing off the latest code.

Also, CI/CD rarely works efficiently in legacy applications because they were not architect-ed to support modern integration patterns. So most companies are either faced with refactoring the system to support a new pattern or they develop a new app and migrate clients.

[Gilletarom]

Originally Posted by rgleason

Continuous Integration ...
1. How to get Beta Testers familiar with Github and locating those compiled files?
2. Are these eventually going to be the source for the Website Downloads page?
....

For almost a month and a half, Did-g has allowed me to have access to his "AppVeyor" relating to its versions of WR, DR and also its versions of OpenCPN ".

So I did "my market" regularly on "AppVeyor" and I realized the beta test. Which is the only thing I can do since I do not know how to develop.

I do not know if it gave him all the help he needed. But, I was able to perform beta tests on versions of WR and DR that Did-g was trying to improve.

- A difficulty, which I encountered, resulted from the identical names given to the different software. One day Did-g did a version of O4.99 and Dave did another one. And after installing Did-g, Dave's one crushed Did-g's.
- Another difficulty, lies in the lack of commentary that can be put in place. Finally, Did-g and I communicated by email. It would be nice if the developer and beta testers on a non-official trial version could comment on "AppVeyor".
- Finally, I had the feeling that Did-g was really annoyed to see modifications of WR, which he had pushed on the GitHub, destroyed by other modifications, pushed on the guitar either by Sean or by other developers.

That said, I had the chance to see WR progressed faster by the possibility of having access to versions of Did-g. But beta testers need to be patient and accept to see some positive developments in the software abandoned along the way.

In any case, I say "Thank you Did-g for this freedom that you granted me".

But how many beta testers could have such freedoms without making the operation too complicated?

Gilletarom.