Opened 5 years ago

Closed 5 years ago

#9783 closed defect (fixed)

gtk-xfce-engine fails with glib-2.54

Reported by: Pierre Labastie Owned by: ken@…
Priority: normal Milestone: 8.2
Component: BOOK Version: SVN
Severity: normal Keywords:
Cc:

Description

See description of the bug at this debian bug report. Note that although the bug talks about gtk2, it is in the gtk3 part of the build...

Change History (8)

comment:1 by Pierre Labastie, 5 years ago

Since this is a problem with encoding (the proposed fix is to transform an iso8859-1 character into utf-8), I am not sure a sed would work. This character is inside a comment containing "Tomas". My proposition is just to remove the lines containing Tomas for the build:

sed '/Tomas/d' -i gtk-3.0/xfce_style_types.h

But I am not sure about copyright issues doing that, since it removes one of the copyright holder name from a file, so making a ticket.

comment:2 by bdubbs@…, 5 years ago

Interesting. There are two different strings:

Tomas Ögren <stric@ing.umu.se>
Tomas Ã<96>gren <stric@ing.umu.se>

Since we are not distributing the file (the user downloads the original), I do not see a copyright issue.

However I did get a sed to fix it. I had a little problem with the Ö, so I needed two substitutions:

sed -i 's/\xd6/O/; s/Tomas ./Tomas O/' xfce_style_types.h

comment:3 by ken@…, 5 years ago

I suppose that the Ö problem was from the locale where you were building it, but anyway the following seems to make the two strings match -

sed 's/\xc3\x96/\xd6/' gtk-3.0/xfce_style_types.h

Not build-tested, I'm back on a machine with glib-2.52.3.

in reply to:  3 comment:4 by Pierre Labastie, 5 years ago

Replying to ken@…:

I suppose that the Ö problem was from the locale where you were building it, but anyway the following seems to make the two strings match -

sed 's/\xc3\x96/\xd6/' gtk-3.0/xfce_style_types.h

Not build-tested, I'm back on a machine with glib-2.52.3.

I'd say it is the other way around, because the python script assumes the encoding is utf8:

sed 's/\xd6/\xc3\x96/' gtk-3.0/xfce_style_types.h

comment:5 by ken@…, 5 years ago

Yeah, I'm in en_GB.UTF-8 and I first looked at the file in view, forgetting that vim has a habit of trying to decide the encoding, and then doing conversions - I should have remembered that, but I piped it to less (and assumed less would respect my environment). When I write it to a file, file tells me it is ISO-8859 text (the original is Non-ISO extended-ASCII text.

With your rearrangement, file tells me it is now UTF-8 Unicode text. Sorry about that, it's a long time since I've seen non-unicode text, except on web pages and Changelogs, and I've stopped worrying about it in Changelogs.

Probably I was also wrong about what sed disliked Ö - it accepted it for me, but that is no longer in my history (I filled that up :-| ) so I can no longer see what I was telling it to do, but probably I changed it to 8859-1 .

For the book, whichever version is used needs -i of course.

comment:6 by ken@…, 5 years ago

Owner: changed from blfs-book@… to ken@…
Status: newassigned

comment:7 by ken@…, 5 years ago

On my system the error came when python2 was used (python3 in debian), but the same error. Before glib-2.54 glib-mkenums was a perl script and obviously it was liberal in what it accepted.

comment:8 by ken@…, 5 years ago

Resolution: fixed
Status: assignedclosed
Note: See TracTickets for help on using tickets.