Bug 17228 - Shorter Compose sequences make longer irrelevant
Summary: Shorter Compose sequences make longer irrelevant
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Lib/other (show other bugs)
Version: git
Hardware: All All
: medium normal
Assignee: James Cloos
QA Contact: Xorg Project Team
URL:
Whiteboard: 2011BRB_Reviewed
Keywords:
Depends on:
Blocks:
 
Reported: 2008-08-20 12:07 UTC by Teemu Likonen
Modified: 2011-10-05 12:13 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments

Description Teemu Likonen 2008-08-20 12:07:25 UTC
Commit a6f4bbf7
    nls (en_US): remove long compositions that override shorter [...]
removed some longer compose sequences because there are shorter ones which take preference over the longer. For example the sequences
    <Multi_key> <apostrophe> <comma> <c>   : U1E09
    <Multi_key> <apostrophe> <comma> <C>   : U1E08
were removed becase there alreade was <Multi_key> <apostrophe> <comma> defined for U201A.

Then commit 4ba09125
    Work on making the en_US and pt_BR UTF-8 Compose as similar as possible
added exactly the same key sequences again. Obviusly they won't work.
Comment 1 James Cloos 2008-08-20 12:23:40 UTC
<mode="brown paper bag">

D’oh!

Thanks for noticing that.

I’ll push out a fix.

</mode>
Comment 2 Teemu Likonen 2008-08-20 12:45:03 UTC
I suggest running a command like

    sed -rn 's/^(<[^:]*>)\s*:.*$/\1/p' Compose.pre \
    | sort >keyseqs.txt

before and after changing Compose files and then comparing the outputs with diff command. Files consist of just compose sequences in sorted order and from the diff one can see if duplicates or conflicting sequences were added.
Comment 3 James Cloos 2008-08-20 14:59:42 UTC
Commit a788792e9de95f8db0639557859722a35087481d repairs that damage.

The same issue remains, however, for pt_BR.UTF-8/Compose.pre, where
it has existed since before the switch to Git.

Any thoughts on whether the U1E08/U1E09 or the U201A sequence should
be removed there?  I have no idea which is more useful for pt_BR users....

(Leaving open pending changes to pt_BR.UTF-8/Compose.pre.)
Comment 4 Jeremy Huddleston Sequoia 2011-10-03 01:19:18 UTC
James, do changes still need to be made for pt_BR.UTF-8?  If so, can you please 
make them or close this bug as FIXED?
Comment 5 James Cloos 2011-10-03 10:49:28 UTC
I think the changes are in.

I’ll confirm and update.
Comment 6 Julien Cristau 2011-10-04 11:44:49 UTC
> --- Comment #4 from Jeremy Huddleston <jeremyhu@freedesktop.org> 2011-10-03 01:19:18 PDT ---
> James, do changes still need to be made for pt_BR.UTF-8?  If so, can you please 
> make them or close this bug as FIXED?
> 
I believe Xlib's make check complains about stuff like that, so this
should be fixed.
Comment 7 James Cloos 2011-10-04 14:23:25 UTC
> I believe Xlib's make check complains about stuff like that, so this
> should be fixed.

Indeed.  It runs libX11/nls/compose-check.pl to do that.

As of my last pull, the only conflict is in iso8859-1/Compose.pre:

Clash with existing sequence in iso8859-1/Compose.pre on line 661: Ctrl<T> <A> <A>
  line #661: Ctrl<T> <A> <A>				: "\305"	Aring
  line #480: Ctrl<T> <A> <A>				: "@"	at

which only gets used if USE_CTRL_SHIFT_T_SEQS is #defined.

Ctrl<T> is used as a poor-man’s Multi_key in that section, so line 480
is a typo; it should be:

 Ctrl<T> <A> <T> : "@"	at

to match the Multi_key sequence in that file.

I pushed a fix for that as f858f3326a.

I beleive this can be CLOSED now.
Comment 8 Julien Cristau 2011-10-05 12:13:36 UTC
(In reply to comment #7)
> I beleive this can be CLOSED now.

Doing so.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.