| 23 Jan 2020 |
| 12:48:51 |  sebl | (i'm also thinking about this from time to time, especially when working on libs or other reusable patches) |
| 12:48:59 | sebl | opinions? |
| 13:35:08 |  ravazquez | looks quite promising, having lunch atm, let me hacve a closer look in a bit |
| 13:37:55 |  azeno | Let me see whether I'll find some time to hack something together for you guys. There is a way to automate it using nunit - which in turn gives nice results on a build server |
| 13:38:27 | azeno | We're doing that internally already - should also be doable from extern |
| 14:02:42 | ravazquez | nice one azeno |
| 14:02:48 | ravazquez | thanks! |
| 15:45:17 |  Hayden_A | @azeno on the subject of testing. are you able to test UI interactions? For instance. what happens if a mouse hovers over a pad, or what happens when an input pin is connected to an output pin, or if a pad is create from a pin? I imagine this is hard to simulate, but probably the place the users come across the most errors. I'm not even sure what the test case would be be, besides 'NoError', But do you think such an automated simulation would be possible? |
| 15:45:45 | Hayden_A | * @azeno on the subject of testing. are you able to test UI interactions? For instance. what happens if a mouse hovers over a pad, or what happens when an input pin is connected to an output pin, or if a pad is create from a pin? I imagine this is hard to simulate, but probably the place the users come across the most errors. I'm not even sure what the test case would be be, besides 'NoError', But do you think such an automated simulation would be possible? |
| 16:17:45 |  tebjan | yes, but the interaction has to be coded. which is quite cumbersome. as you can imagine |
| 16:17:56 | tebjan | In reply to @Hayden_A:matrix.org
@azeno on the subject of testing. are you able to test UI interactions? For instance. what happens if a mouse hovers over a pad, or what happens when an input pin is connected to an output pin, or if a pad is create from a pin? I imagine this is hard to simulate, but probably the place the users come across the most errors. I'm not even sure what the test case would be be, besides 'NoError', But do you think such an automated simulation would be possible? * yes, but the interaction has to be coded. which is quite cumbersome. as you can imagine |
| 16:24:50 | Hayden_A | I bet it is. You'd have to know the position of the pin, code a single click, move then, left, right or middle click. But would this mean you could automate the sreating o f a node and use the .vl document to find the position of pin and execute a tests? |
| 16:26:17 | Hayden_A | I'd personally like to go through a library of nodes and check how each one interacted to check for descrepencies or errors, but of course I wouldn't want to physically do that |
| 21:29:56 | azeno | well well I tried but failed - boils down to our entanglement with .net framework. So will postpone this topic until. Idea was to create a nuget package called VL.TestGenerator which one would simply reference from a msbuild project which in turn would include the vl files to be tested. the package would then have generated the unit tests so they show up nicely in VS as well as can be invoked via msbuild on any build server. |
| 21:30:54 | azeno | So once we move at least our core functionality to .net standard I'll have another look at this - the idea is simple, it's just that those special nugets require .net core etc. etc. |
| 21:36:25 | ravazquez | so close... |
| 21:36:31 | ravazquez | thanks for trying though |
| 22:00:33 | azeno | Well thinking about it maybe I was over thinking it :D well I'll see tomorrow |
| 24 Jan 2020 |
| 08:10:32 |  velcrome | Isn't it just another case where Attributes would solve the problem at its root? |
| 08:13:59 | velcrome | At least from a theoretical perspective- that's how all unit test frameworks I know mark their tests |
| 11:41:48 | azeno | yes would definitely help in marking the test patches, but still, the whole "test runner" infrastructure wouldn't be there |
| 13:15:22 |  iorec | sebescudie: regarding your question: show-help-browser setting is reset sometimes: yes, this happens when you launch an older gamma that does not have this setting yet. this writes the settings-file without that setting and starting the newer gamma then uses the default again. |
| 13:15:55 |  sebescudie | mmmh yeah ok |
| 13:16:31 | sebescudie | let me try again, I'm almost sure I saw it happening when launching only one version |
| 13:16:41 | iorec | ok |
| 13:16:45 | sebescudie | which of course does not sound right |
| 13:18:24 | iorec | ja, let us know if you find something other than we're aware of. |
| 13:21:00 | sebescudie | ok, just rebooted and restarted gamma, no help browser. guess I opened some earlier versions in my previous tests. thanks for the feedback! |
| 13:35:48 | iorec | aight |
| 22:58:36 | |  ada joined the room. |
