On Wed, 24 Jun 2015 12:23:43 +1000 David Seikel <onef...@gmail.com> said:

> On Wed, 24 Jun 2015 11:01:22 +0900 Carsten Haitzler (The Rasterman)
> <ras...@rasterman.com> wrote:
> 
> > On Tue, 23 Jun 2015 12:18:10 -0400 Mike Blumenkrantz
> > <michael.blumenkra...@gmail.com> said:
> > 
> > > I'm fine with this solution, though my preference would have been
> > > to revert it after the internal changes had gone in so that the
> > > errors would be gone for the interim.
> > 
> > i think that at this moment those errors are actually correct and
> > useful. they are saying that the library is broken. it can't adapt to
> > updates on the system and issue events for these etc. without being
> > able to talk to the daemon these features are broken. in fact if its
> > a fresh user then the library will even have no desktop file/icon
> > info because this requires a cache build to be triggered by the
> > daemon. if nothing can talk to the daemon ... then efreet is broken
> > (same for ethumb).
> > 
> > so i see silencing the errors is hiding an issue. when efreet inits
> > its job is to ensure it can see valid efreet related data (desktop
> > files, icons, mime types etc.)
> > 
> > the solution imho is to make sure such errors happen far less often.
> > if we require a dbus session bus then errors will happen a lot
> > outside of a debus session environment. if we move to lower level ipc
> > (ecore_con/ipc) and auto-launching the daemon on demand by the first
> > process that needs it, then these errors will only happen if the
> > daemon is not installed or cannot execute for some reason (an error)
> > or there is a permission etc. issue when it comes to sockets/ipc -
> > also an error.
> 
> Random errors from random programs pop up on dbus all the time,
> invariably crashing dbus itself to the point where you need to log out
> and log in again.

what? i have never seen dbus itself crash. ever. in fact if dbus-launch crashed
enlightenment would go as child processes would go with it.

> Personally I'd go with the "lower level ipc (ecore_con/ipc) and
> auto-launching the daemon on demand by the first process that needs it"
> method.  It's what I used for such things in my own projects.

that is what i am going to get to. i don't buy the dbus instability thing -
never seen this on any distro, but it does add complexity and far more
scenarios which don't work without "active effort".

> > :)
> > 
> > > On Mon, Jun 22, 2015 at 10:11 PM, Carsten Haitzler
> > > <ras...@rasterman.com> wrote:
> > > 
> > > > On Mon, 22 Jun 2015 18:00:38 -0400 Michael Blumenkrantz
> > > > <michael.blumenkra...@gmail.com> said:
> > > >
> > > > > On Mon, 22 Jun 2015 21:12:09 +0100
> > > > > Daniel Kolesa <dan...@octaforge.org> wrote:
> > > > >
> > > > > > On Mon, Jun 22, 2015 at 7:55 PM, Mike Blumenkrantz
> > > > > > <zm...@osg.samsung.com> wrote:
> > > > > > > discomfitor pushed a commit to branch master.
> > > > > > >
> > > > > > >
> > > > http://git.enlightenment.org/core/efl.git/commit/?id=1edb35fff3fe54ac7eea1ba2c26e509284b4e470
> > > > > > >
> > > > > > > commit 1edb35fff3fe54ac7eea1ba2c26e509284b4e470
> > > > > > > Author: Mike Blumenkrantz <zm...@osg.samsung.com>
> > > > > > > Date:   Mon Jun 22 14:53:13 2015 -0400
> > > > > > >
> > > > > > >     edje: unset efreet cache update flag to prevent dbus
> > > > > > > connections
> > > > > > >
> > > > > > >     efreet_init began to be called with the addition of
> > > > > > > translatable strings, but allowing dbus connections in edje
> > > > > > > (and thus edje-derived tools) will result in errors when
> > > > > > > edje_init is called without a dbus connection available
> > > > > > > ---
> > > > > > >  src/lib/edje/edje_main.c | 2 ++
> > > > > > >  1 file changed, 2 insertions(+)
> > > > > > >
> > > > > > > diff --git a/src/lib/edje/edje_main.c
> > > > > > > b/src/lib/edje/edje_main.c index 03c46ea..a891ec5 100644
> > > > > > > --- a/src/lib/edje/edje_main.c
> > > > > > > +++ b/src/lib/edje/edje_main.c
> > > > > > > @@ -1,5 +1,6 @@
> > > > > > >  #include "edje_private.h"
> > > > > > >
> > > > > > > +extern int efreet_cache_update;
> > > > > >
> > > > > > I believe doing this kind of stuff is a very bad idea... I'd
> > > > > > look for a different solution. Arbitrarily externing symbols
> > > > > > like this is something we should never do.
> > > > > >
> > > > > > D5
> > > > > >
> > > > >
> > > > > I'm open to suggestions.
> > > >
> > > > i think the solution is to make efreet not complain so loudly  if
> > > > it can't find
> > > > dbus services.
> > > >
> > > > actually efreet as well as thumb imho create way too  much
> > > > trouble using dbus
> > > > services as an INTERNAL library communication mechanism (between
> > > > daemons and
> > > > library clients). just having to deal with the .service files
> > > > creates a lot of
> > > > user friction - esp those that compile as these files are
> > > > installed in PREFIX
> > > > and dbus isn't looking there. it ALSO means these libraries are
> > > > only partially
> > > > working outside of a dbus session. that makes efreet and ethumb
> > > > useless on systems without dbus or where they may be used on a
> > > > system daemon.
> > > >
> > > > imho for now live with this noise (revert this) and i'll look at
> > > > fixing efreet
> > > > and thumb up to not depend on dbus. debus makes sense when you are
> > > > creating a
> > > > PUBLIC service you expect others to connect to and talk to. for
> > > > an internal feature it makes much less sense given the pain
> > > > points.
> > > >
> > > > --
> > > > ------------- Codito, ergo sum - "I code, therefore I am"
> > > > -------------- The Rasterman (Carsten Haitzler)
> > > > ras...@rasterman.com
> > > >
> > > >
> > 
> > 
> 
> 
> -- 
> A big old stinking pile of genius that no one wants
> coz there are too many silver coated monkeys in the world.


-- 
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler)    ras...@rasterman.com


------------------------------------------------------------------------------
Monitor 25 network devices or servers for free with OpManager!
OpManager is web-based network management software that monitors 
network devices and physical & virtual servers, alerts via email & sms 
for fault. Monitor 25 devices for free with no restriction. Download now
http://ad.doubleclick.net/ddm/clk/292181274;119417398;o
_______________________________________________
enlightenment-devel mailing list
enlightenment-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel

Reply via email to