Posts by John Miles

Google does take down bad ad-links

Several years ago I inadvertently downloaded some malware via a Google ad-link ( a premium line rogue modem dialer - that dates it). In high outrage I complained to Google and within 24 hrs the link had gone. I've not had a similar problem since, but given the scale of their operation and resourcefulness of offenders some must slip through.

Good Article, Bad Language

Interesting and amusing overview of the subject, but why the swearing.

Postfacto selection

One can't simply judge the effectiveness of the process just from the true and false 'detection' rate.

The very fact that the CRB process exists will inhibit a lot of unsuitable people from applying for a position since they know that the CRB check is likely to identify their past record.

This number (whatever it is) would also have to be added to the number actually detected in order to judge the effectiveness of the process in screening applicants - the problem is I doubt that anyone really knows how many people are deterred by CRB.

Single precision float is often enough

It all depends on what you are doing - there are many applications where single precision is sufficient. Or with a bit of thought the computations can be re-arranged to avoid requiring extreme precision e.g. avoid taking small differences between two large numbers.

In the old days when FPUs were about 1/1000 (or less) of todays speed and C would only compile to double precision floating point ops I had to resort to writing a small set of C-callable assembly language single precision routines for vector arithmetic.

It was worth it, reduced run times from 4 hours to just over 1hr !

@Chris Miller - read the article

Whether or not lawyers understand the internet is neither here nor there (and some do very well indeed, see Pinsent Masons 'Outlaw' newsletter).

The point of the legal challenge appears to be to examine whether BSI followed procedures - something lawyers are very adept at assessing. (they are also quite good at reading what an article actually says).

OOXML Voting and Zimbabwe Elections ?

Interesting that the ISO OOXML voting and Zimbabwe election results should both be delayed at the same time.

One would like to think that this is just an accident of timing, and similarities did not extend to dubious vote counting or unexplained about turns in votes, but then again...

Hard part is 100mtr -> 1000 km at 10 and 40Gb.

Going 100m is easy, 50km at 10G has been around for a long time.

The really hard part is getting 10Gb and 40Gb to go further than ~100km.

10Gb pulses of light start to merge together after 100-200km because of dispersion (some components of the pulse travel faster than others).

Up till now you had to put in expensive optical delay compensation filters to correct the dispersion.

At 40Gb the problems get 4 times worse - i.e. the light gets mushed after 25-50km, or you had to go through power hogging electro-0ptics every so often (very un-green).

The Nortel approach cleverly gets round this using techniques similar to those formerly applied to radios and modems - (i.e. Quadrature transmission, dual polarization and line compensation), except scaled up many times to operate at Optical line rates.

Nice camera - but wider zoom would be useful

I'm puzzled why this, like some other 'super zoom' cameras, extend the 3:1 (35-105mm) zoom range of basic compacts at the "long end" only e.g. 35-380mm. (The Panasonic FZ50 is another such example 35-420mm).

Possibly designing a short focus zoom lens is harder than an long focus one, but an effective zoom range of 28-280 or 25-250 mm would seem much more useful.

For the odd shot where one wants something just a bit longer than ~250mm one can always zoom the image by a 20-25% (and make use of those millions of pixels) without loosing too much resolution. But there is no 'post processing' way to get the part of the frame not included in a 35mm shot.