"user profile" in testing

classic Classic list List threaded Threaded
3 messages Options
Terrence Enger Terrence Enger
Reply | Threaded
Open this post in threaded view
|

"user profile" in testing

It is possible to direct LibreOffice to a particular user profile
directory using the command line parameter
-env:UserInstallation=<URL>, and I have grown into the habit of using
a separate user profile for each bug report I look at.

I think that this parameter inhibits the usual conversion of user
profiles across versions of LibreOffice.  Is this right?  Is it a
reportable bug if LibreOffice crashes when using a user profile
created by an earlier version?  ( I tend to the position that every
crash is a bug, but I can well see that a lot of bugs "RESOLVED
WONTFIX" could become annoying. )

FWIW, this question arises from my attempt to reproduce fdo#82722
"EDITING: By editing a macro in the IDE Libo crashes immediatly and
completely when macro opens".  Comment 1 suggests that using a new
user profile makes the crash disappear.  (I have made a SIGABRT
earlier than the reporter's crash, and I have made a SIGSEGV later,
but I have not made a crash where the reporter's crash happened.)

Thank you, all, for your attention and guidance.
Terry.


_______________________________________________
LibreOffice mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/libreoffice
sberg sberg
Reply | Threaded
Open this post in threaded view
|

Re: "user profile" in testing

On 08/17/2014 06:54 PM, Terrence Enger wrote:
> It is possible to direct LibreOffice to a particular user profile
> directory using the command line parameter
> -env:UserInstallation=<URL>, and I have grown into the habit of using
> a separate user profile for each bug report I look at.
>
> I think that this parameter inhibits the usual conversion of user
> profiles across versions of LibreOffice.  Is this right?

No; what happens exactly is:  Whenever at LO 4 start-up no LO 4 user
profile exists (e.g. because -env:UserInstallation points to a directory
that does not exist) and a LO 3 user profile exists at the canonic
location (i.e., not taking into account any -env:UserInstallation) and
the LO 3 user profile does not contain a MIGRATED4 marker file, then the
MIGRATED4 marker file is written into the LO 3 user profile and data
from the LO 3 user profile is migrated into the newly created LO 4 user
profile.

> Is it a reportable bug if LibreOffice crashes when using a user
> profile created by an earlier version?  ( I tend to the position that
> every crash is a bug, but I can well see that a lot of bugs
> "RESOLVED WONTFIX" could become annoying. )

Such crashes should generally be considered bugs that should be fixed.
An exception is when the user profile data that causes the crash can
only have been produced by a "non-official" LO version (e.g. a nightly
snapshot).

Stephan
_______________________________________________
LibreOffice mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/libreoffice
Terrence Enger Terrence Enger
Reply | Threaded
Open this post in threaded view
|

Re: "user profile" in testing

On Mon, 2014-08-18 at 15:58 +0200, Stephan Bergmann wrote:
> On 08/17/2014 06:54 PM, Terrence Enger wrote:
[ incorrect guess about what -evn:UserInstalltion does ]

>
> No; what happens exactly is:  Whenever at LO 4 start-up no LO 4 user
> profile exists (e.g. because -env:UserInstallation points to a directory
> that does not exist) and a LO 3 user profile exists at the canonic
> location (i.e., not taking into account any -env:UserInstallation) and
> the LO 3 user profile does not contain a MIGRATED4 marker file, then the
> MIGRATED4 marker file is written into the LO 3 user profile and data
> from the LO 3 user profile is migrated into the newly created LO 4 user
> profile.
>
> > Is it a reportable bug if LibreOffice crashes when using a user
> > profile created by an earlier version?  ( I tend to the position that
> > every crash is a bug, but I can well see that a lot of bugs
> > "RESOLVED WONTFIX" could become annoying. )
>
> Such crashes should generally be considered bugs that should be fixed.
> An exception is when the user profile data that causes the crash can
> only have been produced by a "non-official" LO version (e.g. a nightly
> snapshot).

Bah!  Humbug!  Despite my best efforts, I am unable to reproduce the
crash which provoked my interest.  It is time to invoke the First Rule
of Holes.

Thank you, Stephan, for your patience with a fat-fingered (and very
frustrated) klutz.

Terry.

>
> Stephan


_______________________________________________
LibreOffice mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/libreoffice