2013/3/20 Marcus Denker <marcus.denker@inria.fr>:
On Mar 20, 2013, at 12:54 AM, Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com> wrote:
First thing, congrats to Pharo community, I observe a good trend. It may seem dumb, but it's great to have pharo executable and pharo icons. Once the image launched, my general feel is good too, though I'm not fan of multiple keys shortcuts. It stressed me to the point that I didn't realize the single key shortcuts were still active in text editor (desperately seeking cmd b+ something... when cmd + shift + N would just browse ivar references). I guess I'll get used to it. Browsing the class Object seems a bit long, I guess there's time wasted scanning the class hierarchy...
Of course, there's a few things to polish. First I ran the tests, and here is what I have to say about it - there were two interactions requests, IMO, there shouldn't be any (a pop up for my name, and a menu)
How to not require a name when a tests does something that needs it? When running tests on the build server it sets a name before and removes it later, but I don't think that the test runner should do that by default.
- there are some failures (10) and errors (3), is this expected? (see attachment)
-> There are some tests that fail when running like this that do not fail on the build server. We should put issues for them on the tracker and fix them.
-> There seems to be a strange Time related bug happening only around midnight. Yes, someone should have a look a that, too. (First step is to submit an issue).
- more annoyingly, during the test, the progressbar feedback disappeared (sporadically flashing) and the TestRunner feedback too (number of succeded test updates went scarce)
ProgressBar still needs to be improved. I think there is even an issue for that.
When I opened a FileBrowser to browse the filedOut TestRunner report, I typed *.txt as a filter and got a RegexSyntaxError nullable closure, that's surprising.
FileBrowser is not used a lot and there are of course no UI level tests. So there will be bugs.
There I stopped exploring the features and inquired a bit about the failures. Most failures are related to DateAndTime. I guess they occured because I was playing with image around midnight with a +01:00 offset.
Yes, two people already saw this bug happening, we should add an issue to the tracker.
In general it just shows: Any issue that is not fixed by someone is not fixed. ;-)
Marcus
Totally agree, but for this specific bug I think that the process can be improved by increased code reviews. The fact that a negative seconds inst var and a julianDayNumber inst var different from julianDayNumber message was chosen indicates that current implementation is a quick hack, a workaround for solving some problems, but clearly not a long term solution. IMO it creates as many problems as it solves. Diff has been added lately, and it is a real progress for reviewing code, but the effort to participate is still a bit high. More eyes, more reviews... The fact that bug only show up around midnight is true only for Europeans which have small offsets from UTC, I guess the probability of bug shall increase in California or Japan. Nicolas