[Pkg-crosswire-devel] Bibledit and Biblememorizer packages

[DM Smith]

On Apr 1, 2009, at 2:14 AM, Eeli Kaikkonen wrote:

> Quoting Matthew Talbert <ransom1982 at gmail.com>:
>
>>>> I don't think waiting on 1.5.12 is a good idea. Right now, Xiphos,
>>>> BibleTime, and Bible Memorizer all work with 1.5.11.
>>>
>>> Well, yes and no -- there are open confirmed BibleTime bugs awaiting
>>> fixes known to be in the SWORD subversion codebase.  Some of those  
>>> bugs
>>> are fairly significant, like making personal notes totally  
>>> unusable on
>>> 64bit Ubuntu.  Probably Xiphos is in a similar position.
>>
>> I don't think that is entirely the case. I believe the bug to be that
>> if you copy the personal commentary from a 32 bit machine to 64 (or
>> vice versa) they will no longer work correctly. At any rate, I don't
>> mean to disagree that some fairly serious bugs have been fixed since
>> 1.5.11.
>>
>
> This 32->64bit commentary issue was reported to BT long time ago. The
> file names are different in 32 and 64 bit platforms. Apparently they
> are generated by the engine using some algorithm which eventually
> depends on architecture. I asked about this in sword-devel but there
> was no answer. This will become more important when people gradually
> move to 64-bit systems.

Eeli,
This is fixed in 1.5.12. I recently found the problem, you then re- 
confirmed it and Scribe fixed it. But it has not been verified that it  
solves your problem.

The problem is that SWORD has a pair of routines to convert from  
little endian to big endian and back. These also constrain numbers to  
particular byte widths. This is supposed to be used when numbers are  
written in binary into files. It wasn't in this case for the Raw Com  
driver, which is used for the personal commentary.

It was supposed to create files with names starting at 1 (maybe 0?)  
and incrementing from there, for each verse being commented on. So if  
someone started commenting with John 1:1, and then John 3:16, these  
would get 1 and 2, respectively. At least that was the theory. In  
reality, they got weird names that worked only on that platform. Going  
from 32-bit intel to 32-bit sun (different endian architectures),  
wouldn't work, nor would going from 32-bit to 64-bit intel (different  
number widths).

These weird names is part of what kept JSword from implementing  
personal commentary. I couldn't figure out the pattern. I let it drop  
and recently came back to it. This time I had an "ah hah" experience  
and verified it was a problem in the SWORD code.

This has been a very long standing, original code bug. It should not  
hold up releasing against 1.5.11.

In Him,
	DM