[bt-devel] [sword-devel] Changes not saved to personal commentary - Possible x86-64 bug in Linux (FC4) - Ref. BibleTime

[Martin Gruner]

Hi Gabriel,

thanks for escalating this.
Of course, if you don't have time then you don't have time. That's how
it is in Open Source projects. Never mind. =) Hope you and your family
are well.

God bless,

mg

am Donnerstag, 24. August 2006 um 05:19 schrieben Sie:

> Hi Sword devs!

> There's a BibleTime bug report that I think may actually be a Sword bug.
>   However, I don't know enough about Sword to even confirm that it is.
> Would anyone be able to help me at least figure out whose bug this is?

> Here's a link to the BibleTime bug report:

> http://sourceforge.net/tracker/index.php?func=detail&aid=1255685&group_id=954&atid=100954

> Here's a brief summary:

> User can write notes to personal commentary, but it's write-once and 
> then read only.  When a verse is edited and saved, it appears to save 
> with no error.  However, when updating the verse the edits do not 
> appear.  Upon inspection, the following anomolies appear:

> 1. The record files in the module's data directory do not have the
>     write flag set.
> 2. Many records appear to have corrupt filenames (e.g. "??" is one
>     file name, "??0000" is another.  Some files have names that are
>     7-digit numbers, some have 14-digit numbers.)

> We've played with umask settings, user permissions, and install 
> locations.  Nothing seems to affect the issue.  AFAIK, nobody can 
> reproduce this bug except the user on x86-64.  However, I know of no 
> other x86-64 testers.  For all these reasons I think it's an issue with
> libsword on x86-64 Linux.

> Vitol statistics:
> OS: Fedora Core 4
> Hardware:  x86-64
> Bibletime:  1.5.2 thru CVS HEAD
> Sword:  1.5.8

> It would be nice if there was a command-line sort of program that ran 
> specific tests on the sword library to test for these sorts of errors 
> (thus eliminating BibleTime as the cause).  However, I don't know enough
> of the Sword API to do this effectively.

> Any help would be appreciated!

> On a realted note.....

> When poking around in the code, I found a few related items:

> 1. The property RawFiles::isWritable() (src/modules/comments/rawfiles/
>     rawfiles.cpp) is based on a cached value, and doesn't change
>     dynamically if the underlying files happen to change.
>     However, this is understandable given the nature
>     of *most* Sword modules.

> 2. The method RawFiles::setEntry() (src/modules/comments/rawfiles/
>     rawfiles.cpp) calls FileDesc::write(), which returns the number
>     of bytes written or -1 on error.  However, the error condition
>     is not checked.  It might be a good idea to check and handle
>     that.

> BTW, I'm very thankful for Sword and The CBS!  You guys are awesome!

> Peace,
> Gabriel


> _______________________________________________
> sword-devel mailing list: sword-devel at crosswire.org
> http://www.crosswire.org/mailman/listinfo/sword-devel
> Instructions to unsubscribe/change your settings at above page



-- 
Mit freundlichen Grüßen
Martin Gruner
mailto:mg.pub at gmx.net