One extended way of testing is using Tor, https://www.torproject.org/download/download-easy.html.en

There is no need to use Tor itself (one can toggle to work without that very slow onion network),
but it comes with a reasonable browser configuration for privacy without too much overhead.

The current version is FF ESR 17.0.8 (that means "Extended Relaese Support" = all security updates
for concurrent FF [= 24.0] are covered, ftp://ftp.mozilla.org/pub/mozilla.org/firefox/releases/
and checked by 'the community').

IIRC it comes without remembering browsing history, disallows 3rd party cookies and cleans up all
the mess (and credentials) on closing the browser. It even does not say 'do not track'.

Be aware that it never should have any plugins (no Java, Flash or Acrobat etc). And it comes with
the addon NoScript, pre-configured (and an almost empty whitelist).


In other words: this provides the basic settings using FF for *some* privacy. No need to use the Tor
bundle, one can do without. Which in no way (to my knowledge) has an equivalent for other browsers.


And a proper implementation should be able to pass that test.

For example: http://www.mapleprimes.com/questions/200021-How-To-Solve-The-Integral--Int2log1sqrts gives the similar errors as described in http://www.mapleprimes.com/posts/200013-Still-Not-Working-With-FF : an error message is displayed for a short time, then it switches to an empty page showing 2013 only.

If I block the scripts from www.mapleprimes.com/omniture then at least it stays with the error message

Edited to add:

It may be a cache problem, FF want's to read from there (which can be seen by
positioning the mouse + right klick  at the "back history" menu bar 1st button),
it starts with wyciwyg, http://de.wikipedia.org/wiki/WYCIWYG, for that hang up.

Also you still have 3rd party cookies (which are not loaded) for that 2o7.net stuff
and if scripts want to use that then they will fail.

BTW: I have DOM storage enabled to its default value, so that can not be the
reason (but 3rd party cookies are disabled, of course)

I think it already was mentioned: after reply one still stays on the
reply - while it already was posted. A bit puzzling

As I see the above mentioned essential calls are now to Mapleprimes, they 
copied sources to their server (thus avoiding calls to external sources).

Thx. I like that.

just to test posting a comment

It simply loads not completely in FF: using addon "NoScript" to control what it allowed the browser addon should show the options which have to be allowed to load the page completely.

It simply does not. Which means: it hangs up before. The server does not send any data which the security addon would reject

And: I am curious whether and when my comments are displayed ...

Ceterum censeo: reset to previous version until the bugs are fixed. It is a pure mess. That community is not around for testing web developments

Also in reasonable settings with FF the complete right 'column' und create post is missing in http://www.mapleprimes.com/recent/all

Inspecting that it seems to have the code

<div id="fb-root"></div>       <!-- SiteCatalyst code version: H.14. Copyright 1997-2007 Omniture, Inc. More info available at http://www.omniture.com -->     <script language="JavaScript"> … </script>       <!-- End SiteCatalyst code version: H.14. -->     <div id="container"> … </div>

It also shows bug.surrogate.27, which I can not find in the page, may be that stems from Ajax (I am not a web coder ... and will stop here)

http://www.mapleprimes.com/posts/200002-Updates-To-MaplePrimes

should have my answer, but does not show it to me ...

I hasitate to write down my very reply I would give a developer about testing ...

http://www.mapleprimes.com/questions/200002-Any-Updates-To-Mapleprimes-Board-SW

@Gaia: For me it is not clear what answer or solution is expected - can you explain ?

May be you sketch in words for what it it is needed and what is the reason for the task.

Also it *might* be, that your floating point number are "too short".

Sometimes it is better to provide "exact numbers", not floats.

May be you sketch in words for what it it is needed and what is the reason for the task.

Also it *might* be, that your floating point number are "too short".

Sometimes it is better to provide "exact numbers", not floats.

@Pavel Holoborodko 

I can believe, that MPFR is faster, and on your site(s) you provide a  C++
lib for that (http://www.holoborodko.com/pavel/mpfr/ but I never used it).

Other fast CAS come to my mind, but PARI is not my very thing.

Also I can imagine that direct implementations for fixed quad double is fast
(I think Bailey has one as well).

So I would be interested in an interface for Maple to use it.

I think it is more easy to study squaring then all the possible
square roots for the task given here.

But can say a bit more, if you wish:

It is more that here I prefer to look at the actual function F instead of some
of its partial inverses invF.

So here instead of looking at invF = sqrt my view is at F: z -> z^2, which is
defined and holomorphic everywhere. 

Finding branch points for some invF = sqrt is the same as finding ramification
points of F (in its image), independend of the chosen invF.

In topology such an F is named a (branched) covering.

Riemann surfaces only come into play if looking at that globally, for example if
one wants to know, how many ramification points exist (like the Hurwitz formula
above, it was over-complicated).

From local complex Analysis one knows, that in any open circle around such a
ramification point the fct can not uniquely inverted: up to certain change of
coordinates it always writes as z^k * unity (this is like a Taylor series,
starting in degree=k, multiplied by an invertible holomorphic funtion).

Now a branch cut is something which has to be removed in the image of F to
allow a partial inversion - which is automatically holomorphic (for sqrt:
no need to choose just the negative reals, it is only a convention).

That means, that even if invF may be extended into that branch cut (sqrt by
contour clock wise around 0) it can not be continuous.

But I guess you already know all this.


The advantage is that one can tread invF = sqrt(z^2 + ...) the same way and
easily finds that F (of degree 4) has a ramification over infinity.

Which was what you pointed out. Now without relying on the FunctionAdvisor
and using exp and log (which would need consideration that their usage will
not cancel out the problem, that is not obvious).


And: sorry for being too complicated in the posts before. Had to re-learn it.

I think it is more easy to study squaring then all the possible
square roots for the task given here.

But can say a bit more, if you wish:

It is more that here I prefer to look at the actual function F instead of some
of its partial inverses invF.

So here instead of looking at invF = sqrt my view is at F: z -> z^2, which is
defined and holomorphic everywhere. 

Finding branch points for some invF = sqrt is the same as finding ramification
points of F (in its image), independend of the chosen invF.

In topology such an F is named a (branched) covering.

Riemann surfaces only come into play if looking at that globally, for example if
one wants to know, how many ramification points exist (like the Hurwitz formula
above, it was over-complicated).

From local complex Analysis one knows, that in any open circle around such a
ramification point the fct can not uniquely inverted: up to certain change of
coordinates it always writes as z^k * unity (this is like a Taylor series,
starting in degree=k, multiplied by an invertible holomorphic funtion).

Now a branch cut is something which has to be removed in the image of F to
allow a partial inversion - which is automatically holomorphic (for sqrt:
no need to choose just the negative reals, it is only a convention).

That means, that even if invF may be extended into that branch cut (sqrt by
contour clock wise around 0) it can not be continuous.

But I guess you already know all this.


The advantage is that one can tread invF = sqrt(z^2 + ...) the same way and
easily finds that F (of degree 4) has a ramification over infinity.

Which was what you pointed out. Now without relying on the FunctionAdvisor
and using exp and log (which would need consideration that their usage will
not cancel out the problem, that is not obvious).


And: sorry for being too complicated in the posts before. Had to re-learn it.