You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The version is numeric in Version.h and labelled everywhere a label is
legal: the installer test parses HMAILSERVER_VERSION with a digits-and-
dots class, VersionInfoVersion is an Inno numeric quad, and the .NET SDK
strips the label back off for AssemblyVersion on its own - so -alpha1
lives on AppVersion, AppVerName, OutputBaseFilename and the seven
.csproj values, and nowhere else. Build 32 rather than 31, because 31
was stamped by 10e885a and seventy commits landed on top of it; a build
number that names a tree which no longer exists is worse than none.
ClassTester's five "hMailServer 6.2.22" literals become "hMailServer
test-generator": they were never version references - a DMARC report
generator argument asserted back out of the XML - and de-versioning them
is what lets the release grep for the previous version come back naming
only history. Deliberately not wired to HMAILSERVER_VERSION; the test
asserts a literal round-trips, and a macro would make the assertion
vacuous.
Both gates ran green today: 1831/1831 on the stamped tree, and
1831/1831 again on this exact tree after the quality fixes landed,
because a run is void the moment anything changes after it. Control
Panel suite 645/645. Schema checkers pass at 76 steps, 54 probes, and a
real SQL CE accepted both the create script and a maximal hm_domains
row - 9,260 declared in-row bytes against the documented 8,060 limit -
so the reviewed row-size concern is refuted by measurement, not
argument.
0 commit comments