Message: previous - next
Month: September 2013

Re: [trinity-devel] (More) Possible bugs introduced from latest k->tde patches

From: "Darrell Anderson" <darrella@...>
Date: Sat, 07 Sep 2013 12:43:45 -0500
On Sat, 07 Sep 2013 02:47:38 -0500 "Fat-Zer" <fatzer2@...> 
>2013/9/7 Darrell Anderson <darrella@...>
>> Additional xsession-error log messages that might be related to 
>> recent renaming (or they might be isolated bugs):
>> * tdeio (KIOConnection): ERROR: Header read failed, errno=104
>> * tdeio (KIOConnection): ERROR: Header has invalid size (-1)
>> * tdeio (KIOConnection): ERROR: Could not write data
>> * tdeio (TDELauncher): ERROR: SlavePool: No communication with
>> slave.
>> * tdecore (TDEProcess): WARNING: _attachPty() 17
>> * tdecore (TDEProcess): WARNING: _attachPty() 13
>> * tdesu: WARNING: unknown super user command
>> For consistency, should KIOConnection be renamed to
>> TDEIOConnection? Looks like tdelibs is the only affected module,
>> although there are some remaining remnant "KIO" usages in 
>> tdepim, tdenetwork, tdebase, tdevelop, tdesdk, tde-i18n, 
>> krusader, kdirstat, tdeio-apt, tork, digikam.
>> Darrell
>Those errors aren't supposed to be something new or related with
>Before patch git hash 5354555 all messages passed to 
>kd{Error,Warning} have
>gone nowhere but now they go to stderr by default.
>So, these is old errors, but previously masked.

Okay, thanks, that explains why I now see the messages when 
previously I did not. Commit 5354555 was pushed to git sufficiently 
close to the k->tde renaming patches. I would not have noticed 
because I was not updating my local source tree during the period I 
helped test the renaming patches with Slavek.

That said, for consistency KIOConnection should be renamed to 

Whether a WARNING or ERROR, the messages are caused by something 
not functioning correctly, especially the ERROR messages.

I understand that commit 5354555 fixed a long standing problem of 
certain messages not being exposed but now that we have that fixed 
we must focus on why the messages are generated. As you shared, the 
messages were always being generated, which means a problem 
somewhere, despite the messages never being seen by anyone.

At this point I now don't know whether the original three messages 
I posted are a result of the renaming or a result of commit 5354555 
now making those messages available and the bugs actually have 
existed for a long time and nobody had any way to know.