[xiphos-devel] attempting to clear matters for a release

Tim Dickson dickson.tim at googlemail.com
Wed Feb 28 17:17:51 EST 2024


Hi Karl,
I made an attempt to use webkit2gtk-4.1 instead of webkit2gtk-4.0 .
editing cmake/XiphosDependencies.cmake changing references to 
webkit2gtk-4.0 to webkit2gtk-4.1 and changing references to libsoup-2.4 
to libsoup-3.0 allows building and running of xiphos, providing yelp is 
built with webkit2gtk-4.1 as well (ie was built with 
--with-webkit2gtk-4-1 passed to configure)

(I do get a few libEGL warnings on the terminal on startup, but if 
people launch from menu or desktop icon, they won't see that)

I did notice that /run/user/0/dconf/user was created out of tree on 
building the package. It would be nice if that could be avoided.

Comments on running xiphos.
Under help, any items that open a browser appear to open chromium, even 
though it isn't my default browser. not sure if that is a xiphos thing 
or something else. The "about" pages all work fine. The "Contents" under 
help is missing. That may well be my packaging. :-(
( xiphos-html-help.tar.gz exists under /usr/doc/xiphos-4.2.1 in the 
package tree )

When running from Konsole, there is a lot of text output to the Konsole, 
such as
Qwidget::paintEngine: Should no longer be called
QPainter::begin: Paint device retuned engine == 0, type: 1
QPainter::translate: Painter not active
QPainter::setClipRect: Painter not active
QPainter::hasClipping: Painter not active
QPainter::setPen: Painter not active
.... lots more similar messages.

aside from that, it runs just fine with webkit2gtk4.1

sorry the git stuff and travis is a mystery to me as well

On 15/02/2024 23:28, Karl Kleinpaste wrote:
> As anyone has seen today who gets issue/source updates from github, 
> I'm doing a few cleanup things toward the goal of a release.
>
> Yes, I've been gone an indecent amount of time again. Regrets and 
> apologies. No, Xiphos has not /quite/ been abandonware, but admittedly 
> it's been close. A couple folks' activities have driven me out of my 
> darkened hovel.
>
> At this time, I need some assistance on a couple of things. Much of 
> the github CI/travis/whatever machinery has always been a mystery to 
> me, others have handled it before, and I'm feeling out of touch.
>
> First, there are 3 open (ancient) PRs and I am trying to determine 
> whether they're relevant.
>
> https://github.com/crosswire/xiphos/pull/1007
> updates USE_GTK_3 to use GTK_CHECKVERSION(3,0,0).
> Fine idea. Seems incomplete. There are a number of instances of 
> USE_GTK_3 not covered by the change. I can always merge it, then do 
> another update referencing it with the remainder, that would be fine. 
> But comments mention that CI is not functional because of repo perms 
> to allow the .com name. I don't know where this is done. Clues for the 
> clueless would be welcome.
>
> https://github.com/crosswire/xiphos/pull/1103
> adds dbus stanza to src/main/CMakeLists.txt
> Seems fine on its face, but comment is that dbus fails on Nix, yet 
> I've never seen a failure of this anywhere. Needed? Critical? Ignorable?
>
> https://github.com/crosswire/xiphos/pull/972
> appears to annihilate GTK2 usage entirely. Sorry, so long as we 
> support Windows build, where we lack adequate GTK3 support (much less 
> GTK4), we are stuck with #ifdef-preserving GTK2 interfaces.
> Am I right about this? Is there better GTK3/4 for Windows?
>
> Other updates already done:
> - Basic copyright update
> - Live chat reference update from freenode to libera
> - WebKit regression workaround using environment variable, since WK 
> people are not fixing that problem.
>
> A main driving reason for all this at this time is that webkit2gtk-4.0 
> is about to go defunct in favor of webkit2gtk-4.1. This will cause 
> trouble as e.g. Fedora 40 and other distros firm up. So updates were 
> merged recently which move us away from 4.0 + soup2 to 4.1 + soup3.
>
> In the CI/travis world, auto-builds are always failing. This hasn't 
> concerned me much up to now but if I'm going to make a release, that 
> stuff ought to run. Aaron Rainbolt made a comment at some point that 
> this is due to a mis-expectation of building in a git tree which ¿is 
> not the case when CI/travis is operating?. Again, I need clues for how 
> to address this.
>
> _______________________________________________
> xiphos-devel mailing list
> xiphos-devel at crosswire.org
> http://crosswire.org/mailman/listinfo/xiphos-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://crosswire.org/pipermail/xiphos-devel/attachments/20240228/762437a4/attachment.htm>


More information about the xiphos-devel mailing list