Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-25 Thread David Faure 
On Friday 22 April 2016 23:51:51 René J.V. Bertin wrote:
> On Friday April 22 2016 23:33:27 René J.V. Bertin wrote:
> 
> > Looks like I'm going to have to add output to kio_http and see if the error 
> > message doesn't come from an error==ENOENT returned from a failed function 
> > call. Seems like something that's going to be fun ...
> 
> Looking at when http.cpp returns ERR_DOES_NOT_EXIST I had a hunch and turned 
> off the web cache. Lo and behold, pages started to load, so the error must be 
> in the caching .

Caching done by kio_http, or caching done at the system level ?

> I keep getting a crash on exist though:
> 
> frame #3: 0x00010ca5bde9 
> libKF5Crash.5.dylib`KCrash::defaultCrashHandler(sig=) + 1049 at 
> kcrash.cpp:527
> frame #4: 0x7fff8b9645aa libsystem_platform.dylib`_sigtramp + 26
> frame #5: 0x00010dc6db50 QtCore`QCache bool>::insert(this=0x7f85eb6f3090, akey=, 
> aobject=, acost=) + 368 at qcache.h:173
> frame #6: 0x00010dc6c6d8 QtCore`setNativeLocks(QString const&, int) 
> [inlined] QStringBuilder::operator 
> QString(fn=0x7f85eb41a500, a=0x7f85eb41a3b0, b=0x7f85eb44e350, 
> rc=) const + 218 at qlockfile_unix.cpp:142

Already fixed in Qt, branch 5.6 (and 5.7)
See comimt b6eea89b67e0d3bb4f8f888fff21257eff0b65a5

-- 
David Faure, fa...@kde.org, http://www.davidfaure.fr
Working on KDE Frameworks 5

___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread René J . V . Bertin 
On Friday April 22 2016 23:33:27 René J.V. Bertin wrote:

> Looks like I'm going to have to add output to kio_http and see if the error 
> message doesn't come from an error==ENOENT returned from a failed function 
> call. Seems like something that's going to be fun ...

Looking at when http.cpp returns ERR_DOES_NOT_EXIST I had a hunch and turned 
off the web cache. Lo and behold, pages started to load, so the error must be 
in the caching .

I keep getting a crash on exist though:

frame #3: 0x00010ca5bde9 
libKF5Crash.5.dylib`KCrash::defaultCrashHandler(sig=) + 1049 at 
kcrash.cpp:527
frame #4: 0x7fff8b9645aa libsystem_platform.dylib`_sigtramp + 26
frame #5: 0x00010dc6db50 QtCore`QCache::insert(this=0x7f85eb6f3090, akey=, 
aobject=, acost=) + 368 at qcache.h:173
frame #6: 0x00010dc6c6d8 QtCore`setNativeLocks(QString const&, int) 
[inlined] QStringBuilder::operator 
QString(fn=0x7f85eb41a500, a=0x7f85eb41a3b0, b=0x7f85eb44e350, 
rc=) const + 218 at qlockfile_unix.cpp:142
frame #7: 0x00010dc6c5fe QtCore`setNativeLocks(fileName=, 
fd=7) + 126 at qlockfile_unix.cpp:160
frame #8: 0x00010dc6be24 
QtCore`QLockFilePrivate::tryLock_sys(this=0x7f85edbd1030) + 612 at 
qlockfile_unix.cpp:190
frame #9: 0x00010dc11870 QtCore`QLockFile::tryLock(this=, 
timeout=-1) + 112 at qlockfile.cpp:212
frame #10: 0x00010c3c0474 
libKF5ConfigCore.5.dylib`KConfigIniBackend::lock(this=0x7f85eb59eb20) + 260 
at kconfigini.cpp:635
frame #11: 0x00010c3a9a13 libKF5ConfigCore.5.dylib`KConfig::sync() 
[inlined] KConfigPrivate::lockLocal(this=) + 275 at kconfig.cpp:109
frame #12: 0x00010c3a9a01 
libKF5ConfigCore.5.dylib`KConfig::sync(this=) + 257 at 
kconfig.cpp:420
frame #13: 0x00010c3a8838 libKF5ConfigCore.5.dylib`KConfig::~KConfig() 
[inlined] KConfig::~KConfig(this=0x7f85eb52b310) + 46 at kconfig.cpp:272
frame #14: 0x00010c3a880a 
libKF5ConfigCore.5.dylib`KConfig::~KConfig(this=0x7f85eb52b310) + 10 at 
kconfig.cpp:269
frame #15: 0x00010bc4aecb 
libKF5KIOCore.5.dylib`KIO::FavIconsCache::~FavIconsCache() [inlined] 
KIO::FavIconsCachePrivate::~FavIconsCachePrivate(this=0x7f85eb52b300) + 118 
at faviconscache.cpp:78
frame #16: 0x00010bc4ae55 
libKF5KIOCore.5.dylib`KIO::FavIconsCache::~FavIconsCache() [inlined] 
KIO::FavIconsCachePrivate::~FavIconsCachePrivate(this=0x7f85eb52b300) at 
faviconscache.cpp:78
frame #17: 0x00010bc4ae55 
libKF5KIOCore.5.dylib`KIO::FavIconsCache::~FavIconsCache(this=0x00010bc83920)
 + 37 at faviconscache.cpp:117
frame #18: 0x7fff80fb37b5 libsystem_c.dylib`__cxa_finalize + 177
frame #19: 0x7fff80fb3a60 libsystem_c.dylib`exit + 22
frame #20: 0x7fff8ad2e604 libdyld.dylib`start + 8

If it were me I'd just disable support for favicons completely but that is 
apparently not foreseen ...

R.
___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread René J . V . Bertin 
On Friday April 22 2016 21:28:56 David Faure wrote:

> Yep. A qt5/kf5 based konqueror isn't released yet, so who would need this 
> yet...

There's a frameworks branch in kde-baseapps that provides "konqueror5", which 
isn't missing too much from what I can see ...

> (dunno about rekonq).

I only

> And only KHTML and QWebkit have KIO integration, QWebEngine doesn't ...
> the browser situation is going downhill :(

I found some mention of a QWebEngine port of kwebkitpart on the webkit ML, but 
not the branch mentioned in that post. Maybe KIO integration is going to be 
included at that level (?)
Or else QtWebKit could become KWebKit now that Qt have ejected it ... could be 
a very good way to attract interest from a wider audience to KF5 ;)

> Set it before starting kdeinit5, and watch kdeinit5 output.
> 
> (or export KDE_FORK_SLAVES=1 to get the kioslave started here rather than by 
> kdeinit5)

Doesn't really tell me much more except that the http kioslave is started:

kf5.kinit.klauncher: KLauncher: launching new slave 
"/opt/local/libexec/kde5/kf5/kioslave" with protocol= "http" args= 
("opt/local/share/qt5/plugins/kf5/kio/http.so", "http", 
"local:/var/folders/j1/1439ppj08xj8h6006s6drbq0gs/T/runtime-bertin/klauncherL41926.1.slave-socket",
 
"local:/var/folders/j1/1439ppj08xj8h6006s6drbq0gs/T/runtime-bertin/kioclientJ41927.2.slave-socket")
kf5.kinit.klauncher: "/opt/local/libexec/kde5/kf5/kioslave" (pid 41930) up and 
running.

Looks like I'm going to have to add output to kio_http and see if the error 
message doesn't come from an error==ENOENT returned from a failed function 
call. Seems like something that's going to be fun ...

R.
___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread David Faure 
On Friday 22 April 2016 20:57:55 René J.V. Bertin wrote:
> On Friday April 22 2016 20:38:09 David Faure wrote:
> 
> >> but the cookies KCM considers that the service isn't running? I'd expect 
> >> something OS X-specific in the KCM if this didn't happen on Linux too.
> >
> >Ah, that was a bug.
> >Fixed -- http://commits.kde.org/kio/9a3ea2c54e8400b29183d2764f39c01d1892dace
> 
> Fixed tonight ... somehow no one noticed this before?

Yep. A qt5/kf5 based konqueror isn't released yet, so who would need this yet...
(dunno about rekonq).

And only KHTML and QWebkit have KIO integration, QWebEngine doesn't ...
the browser situation is going downhill :(

> >Yep, unrelated. You could start debugging with:
> >kioclient5 cat http://www.kde.org
> "The file or folder http://www.kde.org/ does not exist."
> 
> Setting *.*=true in qtlogging.ini doesn't tell me anything more.

Set it before starting kdeinit5, and watch kdeinit5 output.

(or export KDE_FORK_SLAVES=1 to get the kioslave started here rather than by 
kdeinit5)

-- 
David Faure, fa...@kde.org, http://www.davidfaure.fr
Working on KDE Frameworks 5

___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread René J . V . Bertin 
On Friday April 22 2016 20:38:09 David Faure wrote:

>> but the cookies KCM considers that the service isn't running? I'd expect 
>> something OS X-specific in the KCM if this didn't happen on Linux too.
>
>Ah, that was a bug.
>Fixed -- http://commits.kde.org/kio/9a3ea2c54e8400b29183d2764f39c01d1892dace

Fixed tonight ... somehow no one noticed this before?

>Yep, unrelated. You could start debugging with:
>kioclient5 cat http://www.kde.org
"The file or folder http://www.kde.org/ does not exist."

Setting *.*=true in qtlogging.ini doesn't tell me anything more.

R.
___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread David Faure 
On Friday 22 April 2016 18:42:34 René J.V. Bertin wrote:
> but the cookies KCM considers that the service isn't running? I'd expect 
> something OS X-specific in the KCM if this didn't happen on Linux too.

Ah, that was a bug.
Fixed -- http://commits.kde.org/kio/9a3ea2c54e8400b29183d2764f39c01d1892dace

> FWIW, when I try to load a http URL in konqueror5 it tells that there's "no 
> such file or folder" (and that's on OS X only). Probably unrelated even if 
> http and cookies usually go together ;)

Yep, unrelated. You could start debugging with:
kioclient5 cat http://www.kde.org

-- 
David Faure, fa...@kde.org, http://www.davidfaure.fr
Working on KDE Frameworks 5

___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread René J . V . Bertin 
On Friday April 22 2016 15:28:35 David Faure wrote:

> It is, but with a different name. In KDE4 it's
>   $ qdbus org.kde.kded /modules/kcookiejar
> 
> I moved it to the name "org.kde.kcookiejar5" so that later it can be hosted 
> by kiod rather than kded5, transparently.

Right.

So, any ideas why I do see output like

kf5.kded: Loading second phase autoload
kf5.kded: kded module "konqy_preloader" still uses .desktop files 
("kded/konqy_preloader.desktop"). Please port it to JSON metadata.
kf5.kded: 2nd phase: loading "networkstatus"
kf5.kded: Successfully loaded module "networkstatus"
kf5.kded: 2nd phase: loading "proxyscout"
kf5.kded: Successfully loaded module "proxyscout"
kf5.kded: Successfully loaded module "kcookiejar"


but the cookies KCM considers that the service isn't running? I'd expect 
something OS X-specific in the KCM if this didn't happen on Linux too.

FWIW, when I try to load a http URL in konqueror5 it tells that there's "no 
such file or folder" (and that's on OS X only). Probably unrelated even if http 
and cookies usually go together ;)

R.
___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread David Faure 
On Friday 22 April 2016 15:17:43 René J.V. Bertin wrote:
> On Friday April 22 2016 10:19:35 David Faure wrote:
> 
> > kcookiejar5 is just a client-side tool (*).
> 
> And not the tool that applications use to manage cookies?

Right. Apps use library code, rather than command-line tools.

> > The actual service is provided by a kded module. You can see if it works by 
> > doing 
> > $ qdbus org.kde.kcookiejar5 /modules/kcookiejar
> 
> %> qdbus -qt=qt5 org.kde.kcookiejar5 /modules/kcookiejar
> However, interacting with the .org.kde.KCookieServer methods via qdbusviewer 
> gives me either `Arguments: ""` or `(no arguments)` as a result.

Yeah the `qdbus` tool is limited and only supports some basic types.
Doesn't matter, this proves that the service is running for you.

> I'm not seeing the KDE4 kcookiejar in qdbusviewer, yet that one works as long 
> as kded4 is running. Not using DBus?

It is, but with a different name. In KDE4 it's
  $ qdbus org.kde.kded /modules/kcookiejar

I moved it to the name "org.kde.kcookiejar5" so that later it can be hosted by 
kiod rather than kded5, transparently.

-- 
David Faure, fa...@kde.org, http://www.davidfaure.fr
Working on KDE Frameworks 5

___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread René J . V . Bertin 
On Friday April 22 2016 10:19:35 David Faure wrote:

> kcookiejar5 is just a client-side tool (*).

And not the tool that applications use to manage cookies?

> The actual service is provided by a kded module. You can see if it works by 
> doing 
> $ qdbus org.kde.kcookiejar5 /modules/kcookiejar

%> qdbus -qt=qt5 org.kde.kcookiejar5 /modules/kcookiejar
method void org.kde.KCookieServer.addCookies(QString url, QByteArray 
cookieHeader, qlonglong windowId)
method void org.kde.KCookieServer.addDOMCookies(QString url, QByteArray 
cookieHeader, qlonglong windowId)
method void org.kde.KCookieServer.deleteAllCookies()
method void org.kde.KCookieServer.deleteCookie(QString domain, QString fqdn, 
QString path, QString name)
method void org.kde.KCookieServer.deleteCookiesFromDomain(QString domain)
method void org.kde.KCookieServer.deleteSessionCookies(qlonglong windowId)
method void org.kde.KCookieServer.deleteSessionCookiesFor(QString fqdn, 
qlonglong windowId)
method QString org.kde.KCookieServer.findCookies(QString url, qlonglong 
windowId)
method QStringList org.kde.KCookieServer.findCookies({D-Bus type "ai"} fields, 
QString domain, QString fqdn, QString path, QString name)
method QString org.kde.KCookieServer.findDOMCookies(QString url)
method QString org.kde.KCookieServer.findDOMCookies(QString url, qlonglong 
windowId)
method QStringList org.kde.KCookieServer.findDomains()
method QString org.kde.KCookieServer.getDomainAdvice(QString url)
method QString org.kde.KCookieServer.listCookies(QString url)
method void org.kde.KCookieServer.reloadPolicy()
method bool org.kde.KCookieServer.setDomainAdvice(QString url, QString advice)
method void org.kde.KCookieServer.shutdown()
signal void org.freedesktop.DBus.Properties.PropertiesChanged(QString 
interface_name, QVariantMap changed_properties, QStringList 
invalidated_properties)
method QDBusVariant org.freedesktop.DBus.Properties.Get(QString interface_name, 
QString property_name)
method QVariantMap org.freedesktop.DBus.Properties.GetAll(QString 
interface_name)
method void org.freedesktop.DBus.Properties.Set(QString interface_name, QString 
property_name, QDBusVariant value)
method QString org.freedesktop.DBus.Introspectable.Introspect()
method QString org.freedesktop.DBus.Peer.GetMachineId()
method void org.freedesktop.DBus.Peer.Ping()

However, interacting with the .org.kde.KCookieServer methods via qdbusviewer 
gives me either `Arguments: ""` or `(no arguments)` as a result.

I'm not seeing the KDE4 kcookiejar in qdbusviewer, yet that one works as long 
as kded4 is running. Not using DBus?

> 
> (*) the output from kcookiejar5 --help was confusing, I just fixed it.

Ah, good :)

R.
___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

Re: [KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread David Faure 
On Friday 22 April 2016 09:51:03 René J.V. Bertin wrote:
> Hi,
> 
> I'm installing K*5 in a parallel prefix (/opt/local) that also includes Qt, 
> on OS X and Linux. On OS X the DBus lives in there while on Linux (KUbuntu 
> 14.04) I use the system's DBus which is configured to be aware of 
> /opt/local/share too. This works for just about everything, but I cannot seem 
> to get the cookie handler service to work on either platform. Kded5 is 
> running (and is auto-started by `kcmshell5 cookies`) so that's not the issue. 
> I do notice that kcookiejar5 exits immediately without forking when I start 
> it manually.
> Setting "*.*=true" in qtlogging.ini doesn't give me any useful additional 
> information.
> 
> What can this be? I  presume that this service is functional and that the 
> issue I'm having is thus probably related to installing the way I do?

kcookiejar5 is just a client-side tool (*).

The actual service is provided by a kded module. You can see if it works by 
doing 
$ qdbus org.kde.kcookiejar5 /modules/kcookiejar

(*) the output from kcookiejar5 --help was confusing, I just fixed it.

-- 
David Faure, fa...@kde.org, http://www.davidfaure.fr
Working on KDE Frameworks 5

___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel

[KIO] cookie service and installing in a parallel prefix (/opt/local)

2016-04-22 Thread René J . V . Bertin 
Hi,

I'm installing K*5 in a parallel prefix (/opt/local) that also includes Qt, on 
OS X and Linux. On OS X the DBus lives in there while on Linux (KUbuntu 14.04) 
I use the system's DBus which is configured to be aware of /opt/local/share 
too. This works for just about everything, but I cannot seem to get the cookie 
handler service to work on either platform. Kded5 is running (and is 
auto-started by `kcmshell5 cookies`) so that's not the issue. I do notice that 
kcookiejar5 exits immediately without forking when I start it manually.
Setting "*.*=true" in qtlogging.ini doesn't give me any useful additional 
information.

What can this be? I  presume that this service is functional and that the issue 
I'm having is thus probably related to installing the way I do?

Thanks,
R.
___
Kde-frameworks-devel mailing list
Kde-frameworks-devel@kde.org
https://mail.kde.org/mailman/listinfo/kde-frameworks-devel