[sword-devel] Catholic and Catholic2 versification
dmsmith at crosswire.org
Fri Jun 9 11:36:26 MST 2017
BTW, I’ve shortened this, removing country info, so you can reply to it without having it bounce for being too big…
> On Jun 9, 2017, at 2:13 PM, DM Smith wrote:
> Here are the stats for as far back as we have them. Total by module broken out by “agent” and then by country. Other (SWORD) is the installmgr that doesn’t further identify itself like xiphos and PocketSword do. FTP & HTTP - Direct downloads are all downloads that are not identified as one of the others.
> Given the high number of times that PorCap has been downloaded, I think we need to try to maintain backward compatibility.
> [dmsmith at www bin]$ ./moduleStats.pl -name KonNYM,PorCap,SloStritar -start 1900-01-01 -end now
> 16085 PorCap
> By agent
> 8441 AndBible (JSword)
> 3290 Other (SWORD)
> 1490 Xiphos (SWORD)
> 1205 PocketSword (SWORD)
> 1034 HTTP - Direct download
> 277 Other (JSword)
> 251 Search engine
> 97 FTP - Direct download
> 1427 SloStritar
> By agent
> 799 Other (SWORD)
> 196 AndBible (JSword)
> 160 Search engine
> 95 HTTP - Direct download
> 71 FTP - Direct download
> 71 Xiphos (SWORD)
> 21 Other (JSword)
> 14 PocketSword (SWORD)
> 919 KonNYM
> By agent
> 557 Other (SWORD)
> 123 Search engine
> 72 HTTP - Direct download
> 51 AndBible (JSword)
> 49 Xiphos (SWORD)
> 44 FTP - Direct download
> 14 Other (JSword)
> 9 PocketSword (SWORD)
>> On Jun 9, 2017, at 1:33 PM, DM Smith wrote:
>> I’d like to have Troy and Peter’s guidance on this. It is both a software and a module management issue. Back in the day, we knew this was likely to happen.
>> I’m still gathering stats. It is a slow process. I think it should help inform a decision. The upshot is that there are thousands of downloads of these three modules since they have been introduced. Some are more popular than others. Details to follow.
>> The question is how to go forward. One way is to break backward compatibility. The other is to try to create a reasonable migration path. (Which is what I recommend.)
>> From an engineering perspective, we can have any number of versifications in the engine and they can have any name. However, we’ve been reluctant to add versifications, especially if they are one offs. So far versification names are a blend of A-Za-z0-9, I don’t know if there are other restrictions. It seems to follow C variable naming restrictions.
>> Things to consider when updating a module’s conf.
>> MinimumVersion. When we add a new versification, it is tied to a particular release. This means that new module potentially won’t be available until the SWORD engine is released (any day now….) and the front-end is updated.
>> Obsoletes. For a module to obsolete another, it needs a different [Name] and have Obsoletes. It probably should have a different Description (old JSword bug, long since fixed, doesn’t like it when two modules have the same Description).
>> Versification. It should give the appropriate v11n.
>>> On Jun 9, 2017, at 11:13 AM, Cyrille wrote:
>>> Not a lot! And Konnym doesn't have (not yet) the Old Testament. I manage it...
>>> Then if you prefer to write a new v11n, I can do it. But what name to give it? And I would to write 2 v11n, one for esther 10 and one for Esther 16.
>>> A name can be CatholicEs10, and CatholicEs16?
>>> Le 09/06/2017 à 15:46, DM Smith a écrit :
>>>> On the CrossWire server the following are in the main repository:
>>>> Checking the download stats for these. Will get back with the results.
>>>> In Him,
>>>>> On Jun 9, 2017, at 10:37 AM, Cyrille wrote:
>>>>> What the solution proposed? Is it possible to know exactly how modules
>>>>> are compiled with the Catholic and Catholic2? I'm interesting to know it...
>>>>> Le 06/06/2017 à 14:50, DM Smith a écrit :
>>>>>> If an existing v11n changes, all modules released by it will have to be re-released with the update and all front-ends will have to be recompiled with it. JSword will need to change too and front-ends based upon it, also.
>>>>>> The reason is simple the index file for a testament is an indexed array with each verse in the v11n having a predetermined index.
>>>>>> If no modules have been released with it, it is a fairly safe change. The risk are older front-ends with the wrong tables.
>>>>>> Issuing a v2 (simply another v11n with a new name) and deprecating the older probably is a safer change.
>>>>>> In Him,
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the sword-devel