Hey guys. Thanks for everyone speaking on this thread. Andrew, I do appreciate your offer, I sympathize with your single point of failure and widest distribution possible points; however, I tend to agree with the replies given on this thread. The primary benefit I see for a mirror would be faster downloads. I'm not sure we have an issue currently.

Technically:

Changing InstallMgr to 'merge' modules from multiple repositories would be simple to do. This is already done for the concept of global modules typically located someplace like /usr/share/sword/ and personal modules located at ~/.sword/. The method to call is:

        /**
         * Adds books from a new path to the library
         * @param path the path in which to search for books
         * @param multiMod whether or not to keep multiple copies of the same 
book if found in different paths
         *              default - false, uses last found version of the book
         */
        virtual void SWMgr::augmentModules(const char *path, bool multiMod = 
false);


This would allow a frontend to give a unified list of modules from all repositories and would show the latest version available on any repo. I prefer having the repositories separate for the user because it gives a certain flavor/theme/endorsement to the modules. e.g., NET work comes directly from bible.org. Hopefully soon, the ESV will come directly from Crossway, and my hope is that this will become a trend. The Xiphos repo hosts brave modules which give newer features, but which might not yet be fully tested on all platforms (I hope this is a fair characterization).

The mirror concept is different from this identity separation, I realize, but technically, the merging of modules from various mirrors could use the same facility. We would need to tweak the code slightly to add an entry in the memory stamp of the loaded .conf to include all the mirrors which provide the module, so the frontend could offer a choice to the user and then get back to the correct repo to actually request the download, but it would be a simple addition and one a frontend could already do easily if they wanted to offer such.

Also, technical comment to another item on this thread:
We do support multiple copies of the same modules, as seen by the final parameter of the above method. The default is false (don't keep multiple copies), but this functionality was requested by the Bibletime team a number of years ago, and I believe they turn it on by default (or at least they used to). I only saw the benefit for developers, as Karl and David mentioned, so I left the default false.

In general, please don't think distribution isn't one of our highest priorities. My entire push to keep the definition of a 'module repository' as simple as any installed set of SWORD modules, is fuelled directly from that desire-- to technically enable anyone who installs a set a modules, to then become a distributor of those modules simply by making their storage location available to others. If I thought sanctioned mirrors would legitimately increase Bible distribution, I would endorse the move.

Hope this helps,

Troy





On 07/29/2012 05:07 PM, Greg Hellings wrote:
On Sun, Jul 29, 2012 at 8:20 AM, Andrew Thule <thules...@gmail.com> wrote:
On Sun, Jul 29, 2012 at 7:29 AM, Peter von Kaehne <ref...@gmx.net> wrote:
I really don't see your point about this.
To second that - there is essentially no point.
Many publicly available repositories replicate themselves.  (Take
Sourceforge for example).  There are many reason why they do this.
Lower latency in downloads for example, fewer router hops between
client and source download, load balance downloads by distributing
them across multiple sites, reduce available from single point of
failure, and yes even security.  There are places in the world that
filter or monitor sites because of the word 'cross'.  The point should
be obious, suggesting Crosswire consider replicating itself (the site)
and it's module repository has benefits such as no single point of
failure etc - that is unless the chief concern is not text
distribution.
Many do. We have opted not to. The extra technical requirements to
setup mirroring and ensure it stays synced to the master are not worth
the increased benefits to us at this time. If it becomes beneficial in
the future then it will be considered. But at this point the costs
outweigh the benefits, in the group's collective mind.

People who access us from countries which control their internet and
want to block the Bible, need to be cautious and come through proxies,
tor or whatever or obtain stuff via CDs, USB sticks etc.
This might be your preference, but should Crosswire really control
it's own module distribution once it makes them available?
Uhh... yes? CrossWire should definitely control its modules. I'm not
certain what you're getting at here. CrossWire will license any
original works it creates with a very permissive license - GPLv2 for
most software or similarly themed licenses for texts. Those, anyone is
free to use. When Wycliffe or someone else grants us permission to
create a module from one of their texts we negotiate the most
permissive license they are willing to allow. We have no control over
those modules, so there is nothing we can do besides follow the
licenses we were able to negotiate and encourage others to follow them
as well.

Providing secondary download sources in the hope that they will not be
observed while our main ones are - this is silly and actually more
dangerous than going the long way via e.g. tor
Well such a scenario is only one possible benefit.  Tor nodes may also
be filtered (have you ever tried to look at a books.google.com book
through the Tor network for example?)   The benefit of Tor networks is
not that that they get around domain-name restrictions, but that they
get around tracing relationships between client and server.  Likewise,
if someone wanted to monitor access to sword modules (technically)
they would need look at but a single site. But again, this is somewhat
of a tangent.

The offer is kind, is appreciated, but is essentially one which makes no
sense to us, increases our opportunity costs and therefore should be
declined.
No worries.  The goal wasn't to necessarily host a crosswise mirror
(though that was the offer) but to to ask Crosswire's philosophy on
module distribution. I wondered about how to provide maximum
dissemination to Sword modules in my possession not in the main repo,
some of which have not had their distribution rights negotiated.  I'd
like to balance licensing restrictions against broad availability.
If you are the copyright holder, you are welcome to place any modules
you create under a repository. You'd even be welcomed to place that
repository in the master list so that applications can automatically
discover it and offer the source to their users. The Xiphos repository
essentially fits this description. It is not officially a part of
CrossWire, but it is listed in the autodiscover repository list and is
maintained by one of the people in this thread to host modules that he
personally wanted to see hosted but which CrossWire didn't or wouldn't
host.

If you are not the copyright holder then you have to obey the
copyright of the source text. For modules like the NIV or the NRSV,
which we'd like to distribute, we have been unable to come to terms
with their copyright holders. So we can't distribute those and,
legally, you couldn't either. The same goes for non-Bible matieral. If
you want to post copyrighted modules in your personal repository

Having them at a single site neither distributes risk, and represents
a single point of failure.  Similarly, if the applications of rights
such as "Copyrighted; Permission to distribute granted to CrossWire"
implies that module redistribution becomes restricted to one site only
- that likely shouldn't be the license attached to the module. (On the
other hand if Crosswire asserted it's right to text re-distribution
though through a sanctioned mirror program - I'd have less issue with
it).
CrossWire can distribute its sources out across mirrors and multiple
sites if it wants to. If we wanted to host the modules off of S3, we
could do that. If we wanted to acquire a second site hosted in London
to duplicate our data as a failover location, we could do that. So
long as the sites are CrossWire, there is no limitation to CrossWire
distributing the modules from only its one server located in Arizona.
However, we just don't see any need (and yes, we understand the
benefits and drawbacks) for distributed mirrors at this time. Maybe
that will change in the future, maybe not. Maybe Troy will jump in at
this point in the coversation and say, "You know what, I think we
should."

But those modules which are licensed for CrossWire's distribution
only, we are no the ones to place that restriction. The copyright
holders have done so. We graciously thank them for being so generous
with their text and attempt jealously to guard the copyright according
to the owner's wishes. We would like that restriction lifted, because
we do encourage people to download modules that are helpful to them
onto CD or USB stick and hand them to friends and family and
co-workers. That could very much be a violation of such a license, if
those people are not part of CrossWire. Thus, we try to avoid that
license if possible, but if that is the only terms under which a
content owner will license their text, we have to abide by that.

--Greg

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


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

Reply via email to