I am not sure what the original code was intended to do, but asserting on it probably would break loading jpegs with dates in the metadata?

And before the Boud's commit, when the patch could apply successfully, it didn't compile on Qt 5.6, which is still the version we use on Linux :(

In D8542#161718, @dkazakov wrote:

As far as I can tell the code is not a noop, it sums up two dates/times. At least that is what it is expected to do. As far as I can guess, the summing up may happen while merging two layers with metadata.

Sure, its original intention would be to sum up two Date/DateTime values, but as I said it does not make sense. Even if the addition result is assigned properly, the result would be meaningless.

And you're right that the code isn't a noop... it's something else: It replaces the value with a default-initialized QDate/QDateTime and completely discards the original values. QDate::fromJulianDay and QDateTime::toSecsSinceEpoch are static functions, which means they don't operate on and cannot change the instance. Some code linting would've caught it.

Anyway, the code is definitely broken in some way. I would still suggest removing the broken addition and just keep the original value, but changing the assert to a warning log message. It looks like the code might not even be called at all according to Boud (see below).

There is also a problem that the patch doesn't apply to master anymore :(

In D8542#161723, @dkazakov wrote:

And before the Boud's commit, when the patch could apply successfully, it didn't compile on Qt 5.6, which is still the version we use on Linux :(

Yeah, Boud fixed the build errors with a series of commits. This patch alone would not have fixed build issues on Qt 5.6. (843d4348989ac1f49f705a8c9b8f98576cdbf1cc is the missing change.)

P.S. My previous irc conversation with Boud:

<boud> windragon: outch, I'll have to ifdef that
[...]
<windragon> boud: does it ever do QDateTime arithmetic?
<windragon> boud: perhaps just change it to a warning log instead
<windragon> boud: idk either there is a bug in the original code and it should store the addition result properly, or that it doesn't ever get called at all
<boud> windragon: yes, it's pretty weird... only cyrille probably can remember how this was meant.
<boud> And I don't know even know when the whole function would be called
<boud> I've pushed an ifdef, but for what I can see, it probably is never called with a QDateTime
<boud> it's called from here: https://pastebin.com/jbYVKAkK
(content:)

mergeEntry(dst, srcs, dcSchema, "contributor");
mergeEntry(dst, srcs, dcSchema, "creator");
mergeEntry(dst, srcs, dcSchema, "publisher");
mergeEntry(dst, srcs, dcSchema, "subject");
mergeEntry(dst, srcs, XMPRightsSchema, "Owner");
mergeEntry(dst, srcs, XMPSchema, "Identifier");

<windragon> I'll change it to a warning log message and update the patch then
<windragon> maybe later
<boud> okay
<boud> I;'m not even sure that "merging metadata" should be expressed as adding.
<boud> of course, Krita might very well be the only application that merges metadata when layers are merged; I don't think any other application keeps per-layer metadata.