Updates from Sine Nomine Software

This weekend we have released some further updates to software, as mentioned at the Wakefield Show.

Impact 3.60 relational database and ImpEmail 1.20 email merge
A new display of field definitions for the database is the main new feature, but there are several other improvements, including better interpretation of dates when just two digits are entered for the year. ImpEmail gains Bcc, Reply-To and Sender fields, and the editor is improved so that triple-clicking selects an entire merge command.
This is a free upgrade for existing users available from the Impact web pages

RiscOSM 2.01 vector mapping software
Version 2.00 of RiscOSM was released a fortnight ago, and this is the traditional post-show bug fix release. Version 2.00 introduced a new map data format and changes which speed up map loading and rendering, particularly for small-scale maps. It is a chargeable update available from the RiscOSM web pages

Version 2.01 adds

• two more small speed enhancements
• corrections to interpretation of normal/daylight-saving time for tracks
• bug fixes, including a crash when saving pins as KML

Recce 1.05
This is a small update to version 1.04 which was released the day of the Wakefield Show and which adds weather forecasts to the data accessible from RiscOSM. Version 1.04 is a chargeable update. See the RiscOSM upgrade page for ordering details.

Important note about CPUClock
We have had several reports of strange errors in RiscOSM over the past couple of months. These have been tracked down to a faulty version of CPUClock which was corrupting the floating point environment. If you have CPUClock 2.08, please upgrade to CPUClock 2.09
Many thanks to Chris Johnson for tracking down this issue quickly.

SuperDoku 1.32
A new version of SuperDoku, the puzzle generator and solver, has received a small update: literally adding an equals sign to one line of the source code. This corrects a bug which meant the software was unable to solve the Hard Futoshiki puzzle in the 23 April Guardian Weekly. Other puzzles may have been affected too. Actually the hints have been improved as well, so it was a bit more than a single equals sign. See our general software page for the download.

CPUClock which was corrupting the floating point environment

Ah. DigitalCD has been throwing up a possibly related error recently. Maybe…

CPUClock which was corrupting the floating point environment

I wonder if this is the cause of NewsHound issue some of us have had, I use CPUClock.

Weird floating point issues certainly afflicted me in NH (plus RiscOSM and other unidentified apps) … and stopped when I installed CPUClock v2.09, but at about the same time I also installed DaveH’s latest NH test version, which removed floating point anyway.

DCD has stopped crashing, using CPUClock 2.09.

How does one kill the AudioMPEG module when it continues playing, without a machine reset?

RMKill?

If DCD dies AudioMPEG continues its feed. The module refuses to die if it’s in use, so no.

Just happened in this instance so not important. Just curious.

If DCD dies, use this command from BASIC:
SYS “DiskSample_Configure”,-1

this will close all open streams which should in turn unlock AudioMPEG, Vorbis and AC3.

For Tracker modules:
SYS “TimPlayer_Configure”,-1

This reminds me that I should maybe add !ResetSnd to the DCD archive, once run it opens a panel allowing to reset different sound modules like DiskSample, TimPlayer, AmPlayer, QTM, DSym, PlayIt.

Version 2.02 of RiscOSM is now up on the web site: a further bug reported by a user was fixed today.

Yes, the problems with NewsHound were certainly to do with CPUClock. Applications using floating point instructions could have been affected by the bug in CPUClock 2.08, as it had floating point code in its module which was corrupting the floating point environment. Dave Higton fixed NewsHound by removing the floating point code from NewsHound, leaving nothing to corrupt. About the same time, Chris Johnson removed the floating point code from the CPUClock module.