Jens,
Thanks for the speedy response. I certainly appreciate the low-bandwidth
situation of the folks who generously contribute their time. Nonetheless, IE is
the browser used by government and many corporations. This is not a matter of
what browser is better than another. It's just a fact of life. I as a
contractor, don't get to dictate the tools used in those environments. IT
staff, like database administrators are absolutely not going to use curl
(assuming it's even allowed to be installed on machines) to manage a database.
CouchDB has to have browser admin UI that is compatible with IE, Chrome and
Firefox. IE being the most important even if is not the choice of developers.
I'm presented with such frustrating realities every day in my work for big
companies and government. I'm sure there are plenty of folks on the user and
developer list who deal with that.
I'm curious as to what the QA process is for posting official releases. What is
the test regression process? I got hung up by a serious bug (which I didn't
know was a bug) last week in the Mac OS 1.5 build that I got off the web site.
I found that modest size attachments (e.g. 3MB) took 20-25 seconds to upload,
regardless of the technique. I ran the same test on Windows and found no
problems, under a second, barely measurable. One the contributors, who respond
to my post on the user forum, said the bug been identified and fixed in a
branch, but not yet posted as a 1.5 patch build. The Mac is not my production
platform, but I develop all my code on it. Do you know when/how the bug fix (by
the way, what is the bug) will be posted to an official build?
Thanks
Jason
On Feb 18, 2014, at 11:04 AM, Jens Alfke <j...@couchbase.com> wrote:
>
> On Feb 18, 2014, at 10:11 AM, Jason Winshell <ja...@jasonwinshell.com> wrote:
>
>> I'm at a bit of a loss trying to understand how a bug like this could go
>> undetected in an official release.
>
> Because the CouchDB contributors don't have the bandwidth to test the admin
> UI (which is obviously lower priority than the core database) on every
> version of every browser?
>
> (Also, I'll bet there's a low correlation between people who work on or use
> CouchDB and people who use MSIE, so this kind of bug wouldn't get noticed
> very often.)
>
> The thing I find odd is that the one browser CouchDB is officially tested
> with is Firefox, not Chrome, when Chrome has about twice the market share of
> Firefox or MSIE[1] and is based on {something still very close to} WebKit,
> which has about 80% of mobile market share[2].
>
> I'm guessing the choice of which browser to support/test with was made early
> on, maybe 5 years ago, when Firefox was more dominant and MSIE was still in
> terrible shape.
>
>> It's essential that Futon work with IE. Corporate and government
>> installations dictate the what software is installed on machines. A user
>> would not be allowed to install a Couch compatible browser.
>
> Futon's not an essential part of CouchDB. It doesn't do anything you can't do
> through the REST API. As long as users are allowed to install 'curl' they'll
> be OK ;-)
>
> —Jens
>
> [1] http://en.wikipedia.org/wiki/Usage_share_of_web_browsers#Summary_table
> [2] http://www.businessinsider.com/mobile-browser-share-2013-11
