On Tue, Feb 20, 2018 at 8:28 PM, Scott Kostyshak <skost...@lyx.org> wrote:

> On Wed, Feb 21, 2018 at 02:17:33AM +0000, Joel Kulesza wrote:
>
> > Understood.  I suppose one aspect of "emphasizing" that was veto'd was to
> > include a note that the main download site is based in France.  This
> seems
> > like an odd omission to me because all mirrors have their location listed
> > and because the main download site's URL doesn't give an indication of
> > location naturally (e.g., not a .fr or similar TLD).
>
> It seems irrelevant to me where the server is. The server could be in
> the U.S. and just be very slow. Or, the server could be in France and
> could work just fine for someone in the U.S. (as it works for me).


True.  However, my thought was that based on the poll, for every response
that tested both the default and UCSD, UCSD was faster.  Because no average
speeds are posted (which is reasonable because of the large variation and
unpredictability) and because locations are only posted for mirrors, if I
experienced slow downloads from the main site there is nothing that would
prompt me to try another thinking it was closer/faster/etc. other than
curiosity.  Were I to know the main download link is not physically near me
versus another that is clearly so (and likely then taking fewer routing
hops), I'd be more inclined to try a mirror if I experience slow speeds.

Naturally, I'm making the case on account of others.  I've adjusted my own
behavior to pull from UCSD; however, it's not clear to me that new users
would explore mirrors trying to optimize their retrieval.  Further, I
(perhaps unfairly) generally regard mirrors as a secondary option to the
main site assuming that site will be fastest/most
stable/auto-negotiated/etc.


> What
> is relevant is whether the server is fast enough for you, and you can
> only see by trying it.


True, but again, using distance as a surrogate to routing hops can give a
clue, but that metric is only available if full information is given for
all downloads sites, not just auxiliaries.


> But if another developer gives a +1 for
> specifying that the server is in France, I will listen.
>
> In any case, this discussion on server speeds is a good one to have and
> I'm glad we collected a poll and made progress on making it better.
> Thanks for all of the suggestions.
>

I'm also happy to have the discussion and hopefully to make it most clear
to a new LyX user where they can most quickly obtain the program.

Thanks,
Joel

Reply via email to