[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-mac
Subject:    Re: [KDE/Mac] Multiplatform frameworks
From:       René_J.V. Bertin <rjvbertin () gmail ! com>
Date:       2015-03-02 8:58:07
Message-ID: 5487059.KDnyd4PXlF () patux
[Download RAW message or body]

On Monday March 02 2015 08:16:14 Marko Käning wrote:

> I remember to have had discussions about these locations with David a while ago
> on KDE-MAC…
> 
> I recall him also pointing out that some apps might not be KDE-based, but using
> KF5 libs back then, which complicates things even more.

Yeah, "funny" how without him participating in the discussion things centre back \
around KDE (or should I just say "K"??) ...


BTW, I've added my €0.02 to https://bugreports.qt.io/browse/QTBUG-44473, which an \
updated version of the patch that's currently in my qt5-mac-devel port.

> all on my own. As you know, I finally had to come up with an OSX/CI system which
> uses almost the same folder structure like KDE's Linux build slaves. (Except the
> application install location in /Applications/KF5/ everything looks like on Linux.)

Well, that might be taken as either one of 2 indications ... 1) we should use linuxy \
paths or 2) screw CI ...

> I think I saw a ReviewBoard item where the path had to be encoded, so the
> space became %20 (or some such).  Are you expecting anything worse?

I'm expecting to run into the kind of problems where in the end you learn that the \
app developers simply don't support paths with spaces in their name.  Introducing a \
quoted-printable encoding isn't really going to help those applications (rather than \
getting errors you might find your filesystem littered with entries showing someone \
took the encoding verbatim).

The fact that there's no good reason this would happen doesn't mean it won't.

> Yes, that is some strange puppy: a private constructor, but no
> implementation and no instantiation anywhere.  And QSP has no data-state
> ATM, static or otherwise, except what is on the stack.

A private ctor? I can't recall having seen one at all. QSP *seems* to be "just an \
exercise" in using OOP techniques where simple functions would have been enough. No \
offence intended, David; I'm sure there's a bit more to it than that.

Anyway, it wasn't very hard to implement a 1st draft of my ideas; see \
https://bugreports.qt.io/secure/attachment/46944/QSP-patch-20150301.diff . There's no \
ctor, but there is only 1 function that needs to decide according to what guidelines \
to construct its return value, at least for now.

This is currently in testing on the pixilla VM, including the question "what happens \
when you call the function with the new argument from code that wasn't recompiled \
with the new header declaring the variable and its default value" :))

BTW, we'd need to discuss whether ${prefix}/config is a good choice for the CONFIG \
locations; Linux uses /etc/xdg for that and we do have ${prefix}/etc/xdg ...


R.
_______________________________________________
kde-mac@kde.org
List Information: https://mail.kde.org/mailman/listinfo/kde-mac
KDE/Mac Information: http://community.kde.org/Mac


[prev in list] [next in list] [prev in thread] [next in thread]