| broulik |
it doesn't have technical downsides but ends up creating urls of the type
scheme://@host/path
which is technically equal to scheme://host/path, so the excess @ is really
just not necessary. simply check if the username is empty and if so
do not set a username on the QUrl
| Automatic diff as part of commit; lint not applicable. |
| Automatic diff as part of commit; unit tests not applicable. |
I wonder why QUrl doesn't clean that up on its own
| knetattach/knetattach.cpp | ||
|---|---|---|
| 204 ↗ | (On Diff #43798) | lazy, const QString |
I think ... since this gets put into the desktop file via toDisplayString and toDisplayString has the behavior that
With the default options, the resulting QString can be passed back to a QUrl later on, but any password that was present initially will be lost.
what QUrl does is probably accurate. In that an empty QString is not a null QString, so if the goal is to reconstruct the same QUrl out of the string then encoding the empty username like that is reasonable. I've had a quick grep through rfc2396 and from a spec POV @host == host though.