286 reputation
1210
bio website mnapoli.fr
location
age
visits member for 4 years, 3 months
seen 2 days ago

I'm a software engineer passionate about coding, web development and applications architecture.

You can read more about me on my blog or on my GitHub profile.

Here are some projects I'm working on:

  • PHP-DI - Dependency Injection library for PHP

  • Couscous - Put your Markdown documentation online

@matthieunapoli


Feb
11
comment Alternative to “Passing/Broken build” indicator?
@Earlz Exactly! What I'm wondering is "is it done by someone, somewhere? And are there tools that support that (CI and unit testing libraries?" Because if I just categorized those tests with classic CI and unit-tests tools, the build will always fail anyway and I won't see a difference between which tests failed, and so it won't be useful :/
Feb
11
awarded  Nice Question
Feb
8
comment Alternative to “Passing/Broken build” indicator?
Simply tracking the number of failing tests is insufficient that's not the only possible metric. For example: Branch-A improves it to 40 tests (80% passing) with no regression. No regression means previously passing tests always pass. In short, a test would be allowed to fail as long as it has never passed. It seems to me that we are missing on good things by constraining to forbid failing tests in main branches. (of course it would require tools working differently: unit tests, CI, ...)
Feb
8
awarded  Commentator
Feb
8
comment Alternative to “Passing/Broken build” indicator?
@MadKeithV Exactly
Feb
8
comment Alternative to “Passing/Broken build” indicator?
the build would always have failing tests precisely! But is that such a bad thing? Our only metric is "the build is broken or not", but your code could be full of known bugs, so that doesn't really mean anything except there is no regression. In a perfect world, every tracker issue would have a test (reproducing is easier than fixing). So the upside would be to see that 35 tests/70% of all the tests are passing, that Branch-A improves it to 40 tests (80%) with no regression, and that Branch-B has regressions. Today you could only say Master and Branch-A are OK and Branch-B is broken.
Feb
8
awarded  Editor
Feb
8
revised Alternative to “Passing/Broken build” indicator?
added 9 characters in body
Feb
8
comment Alternative to “Passing/Broken build” indicator?
Indeed you make a point with topic branches. But I'm never talking about commiting broken code, just failing tests. For example I could help an open source project by creating tests for incoming tickets even if I don't know how to fix them. That saves a some time for the maintainers.
Feb
8
comment Alternative to “Passing/Broken build” indicator?
@JoachimSauer Yes but is such metric standardized/used in any major project? I'm trying to understand why most of the projects (and CI tools) work that way.
Feb
8
asked Alternative to “Passing/Broken build” indicator?
Dec
24
awarded  Yearling
Dec
24
comment Presentation VS Application layer in DDD
Thank you, indeed you made me realize that for my case separating Application and Presentation is useless. Simplicity first!
Dec
24
accepted Presentation VS Application layer in DDD
Dec
24
asked Presentation VS Application layer in DDD
Oct
5
accepted Should a domain expert make class diagrams?
Oct
4
comment Should a domain expert make class diagrams?
@DocBrown this is the first case, the UML models are used as "technical specifications", this is why i thought this might not be optimal
Oct
4
asked Should a domain expert make class diagrams?
Aug
7
awarded  Notable Question
Jul
13
comment SVN code review tool
Seem to only work with Visual Studio