ahmadsamir added a comment.
In D24588#546367 <https://phabricator.kde.org/D24588#546367>, @vkrause wrote: > In D24588#546229 <https://phabricator.kde.org/D24588#546229>, @dfaure wrote: > > > Hmm, why not just do `KSslErrorUiData uiData(socket);` in the caller, or even just `askIgnoreSslErrors(KSslErrorUiData(socket), storedRules)`. > > Don't know, just wondering (not objecting). @vkrause, any opinion? > > > That would be my gut feeling as well, ie. have sslui depend on the abstract error representation, not on all possible error sources. OTOH it's not changing the dependency situation overall here so no strong objection to that either. In any case the KTcpSocket variant should be deprecated eventually. That makes more sense, (so I take what I said previously back... partly anyway; this way the implementation detail, i.e. the socket type, is hidden from SslUi so that it needs only a KSslErrorUiData). REPOSITORY R241 KIO REVISION DETAIL https://phabricator.kde.org/D24588 To: ahmadsamir, dfaure, vkrause Cc: kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, bruns
