Closed ThomasBreuer closed 2 months ago
The test failures are not due to the proposed change.
The error message Error, Variable: 'DatabaseAttributeAdd' must have a value
suggests that the recent upgrade from CTblLib 1.3.7 to 1.3.8 is involved in the problem, because the new version of CTblLib introduces DatabaseAttributeAddX
instead of DatabaseAttributeAdd
, in order to make some functions of CTblLib independent of the Browse package; the function DatabaseAttributeAdd
belongs to Browse.
I do not understand why the current error did not occur in the tests before CTblLib 1.3.8 got accepted, see gap-system/PackageDistro/pull/920.
See also #5672, I've encountered the same issue.
For that: Can we assume that the system program rev is always available?
Of course it's not always available everywhere, almost nothing is. But it is available in the CI environment were we run those tests, and we also already require it in tst/testspecial/child-process.g
, so I think it is safe enough to expect it.
Up to now, the example for
InputOutputLocalProcess
promised a string of length 4095, but 10002 is more reasonable.Can't we turn this example from
<Log>
(which is not tested automatically) to<Example>
? For that: Can we assume that the system programrev
is always available? If yes then all we have to change is to hide the local paths that occur in the example, if not then we should invent a different example.