Archive Magazine CD 1995/Archive Magazine CD 1995.iso/text/hints/volume_05/issue_12

Text File | 1995-02-16 | 9.8 KB | 215 lines

Hints and Tips
5.12
•   Free Space − (Ref to Archive 5.11 p21) − The call to check the amount
of free space on a disc is ADFS_FreeSpace. On entry, R0 points to the
disc specifier and on exit R0 contains the free space and R1 contains
the size of the largest object that can be created. E.g.
5.12
SYS“ADFS_FreeSpace”,“:0” to free%,largest%
5.12
Change the ADFS to RamFS, SCSIFS or whatever you’re using.
5.12
•   Free space on hard discs − For some time I had been concerned by the
apparent discrepancy between the figures returned by COUNT and FREE for
the space used by files on my SCSI discs. I was aware that COUNT only
counts the actual files and that FREE counts all the space used but I
still felt uneasy and my discs seemed to fill up alarmingly quickly.
5.12
ADFS floppies use 2k for every directory and there is an initial 4k
overhead for the map on E format discs. My impression was that the same
amount of space would be used for each directory in all Filecore
systems, including SCSI, but that the map would take more space
depending on the size of the disc.
5.12
So the first thing I did was to estimate the number of directories on my
120M disc. I make full use of the directory structure to organise my
data, not to any great depth but with some width. RISC-OS encourages
this and, of course, applications are directories too. I reckoned there
were well over 1000 directories on the disc, but at 2k per directory,
this did not account for the difference.
5.12
With a little experiment of adding a directory and looking for the
change in free space, I found that a directory was taking a staggering
28k! This was where the space was going, but why?
5.12
My Oak SCSI card is managed by their SCSIForm program which allows some
flexibility in defining the format of the disc. Having backed-up, I did
some experimenting with the parameters.
5.12
I changed the sector size, but the minimum value for the 120M disc was
512 bytes and, for a 42M removable was 256 bytes. I imagine the lowest
possible value of 128 bytes is only for small discs. Anyway, the changes
I made had only very small effects on space or disc transfer times. The
default is 512 bytes per sector, so I left it at that.
5.12
Next I changed the file allocation size. This defaults to 2k on the
large disc and 1k on the other. The minimum is 256 bytes. The larger
values are chosen, I believe, to speed up transfer times to large files
on larger discs. I did not think this parameter would have an effect on
the structure, but look at the table below. The results were impressive.
For a loss of transfer time of less than 5% I was gaining a saving on
structure of 7 times! There were similar results for the smaller disc.
5.12
   File Alloc’n  (bytes)   2048
   1024   512   256
5.12
   Map Space (k)   57   47
   68.5   124
5.12
   Dir’y Space  (k each)   28   15
   7.5   4
5.12
   Mode 0   (k/sec)   1959   1864
   1864   1864
5.12
   Mode 21  (k/sec)   994   955
   955   955
5.12
I have not delved any further into why more than 2k per directory is
taken in SCSI but the figures suggest 14 blocks are used: 13 are <256
bytes and the other is 512-768 bytes.
5.12
When I settled on a file allocation size of 256 bytes, I recovered about
30M of extra space on the larger disc − about £200 worth, perhaps. Some
of this saving comes from taking much less space for the multitude of
small !Run,!Boot and !Sprites files.
5.12
The only other make of SCSI of card I have access to is a Cumana one in
school, running in a fileserver for 600 users. This too has 1000+
directories. The Cumana SCSIManager program only offers formatting with
default values, so I changed the format using my Oak card and made
similar impressive savings. The disc runs happily with the Cumana card,
as you would expect, but will not dismount. This is no penalty with
self-parking heads.
5.12
I then found an old ADFS hard drive and did the same check, finding that
a new directory takes 12k. RAMFS takes a straight 2k. I would be
interested to know if the same problem applies to IDE discs and what
formatting is possible with other SCSI cards.  Steve Drain, Portland.  A
5.12




Impression Hints & Tips
5.12
Ŷ•   Problems with version 2.17 − If you are having problems with version
2.17 − such as the machine hanging up when you select large areas of
text − don’t blame Impression! Basically, what is happening is that CC
have implemented message-passing via the Wimp for their thesaurus and
these messages are rather large. This is a perfectly valid thing to do
under Acorn’s RISC-OS programming guidelines but not all filing systems
have been written to allow for such large messages and they end up over-
writing areas of memory − hence the crashes. Filing system versions
known to be a problem are:
5.12
   Oak SCSI version 1.11
5.12
   Ian Copestake IDE version 1.13,
5.12
   Serial Port IDE version 1.01, March ’91
5.12
CC say... “If your Archimedes gives problems, contact your Acorn
supplier or the appropriate company. Please do not contact Computer
Concepts − these problems are not caused by Impression but by the filing
system.”
5.12
The Oak SCSI filer 1.11 has been replaced by version 1.16 but this ‘new’
version (actually two years old!) is available through Archive as an
upgrade for £12 inc VAT.
5.12
The ICS filer 1.13 can be swapped for version 1.14 as a free EPROM
exchange from ICS. They also have a completely re-written IDE filer
(version 2) which is available from them for £15 +VAT. The customer
returns the old EPROM afterwards.
5.12
The Serial Port filer 1.01 can be replaced with 1.02 as a free upgrade −
in this case, just send the floppy disc back to The Serial Port and they
will replace it.
5.12
• Using borders − From David Wooldridge comes one of those tips that is
obvious to the initiated but might be new to some users. When choosing a
border in the Alter Frame window (view mode), don’t forget you can see
the various borders by pointing at the white border in the window and
pressing menu. A border picker window opens, allowing you to select from
all the available borders. This tip is in the manual but, as we all
know, not everyone reads manuals − own up, please, if this is new to
you! (Me for one! Ed.)
5.12



The Readers Write!
5.12
• Laser Jet printing with RISC-OS3 − Gerald Fitton (of Pipeline fame!)
has experienced a problem when using RISC-OS 3, !Printers, Laser Jet II
and Impression. He has found that documents in the portrait format print
7 mm out of position on the vertical axis and are offset 5mm to the
right. This occurs even when the page borders are set to 0 mm. If he
increases the latter figure, the print out is further out of position.
As Gerald says, ‘What I need is a minus value for my borders. Help!’ Has
anyone else come across this particular problem?
5.12
• Keeping track of Draw and Sprite files − In Archive 5.9 p37 Steve
Hutchinson was asking advice on keeping track of his Draw and sprite
files and was recommended two viewing applications. There is also an
easy hardcopy method namely !PicList which was on the Acorn User October
1991 disc after being upgraded from the May version. The application
makes a set of sprite “pages” containing any Draw or sprite pictures
that are dropped into it. The pictures are displayed in a grid and are
labelled with their name/path above them. The application is easy to use
from the iconbar having options to:
5.12
•   Set up the size of the “page” (defaults to print size if a driver is
loaded),
5.12
•   Set the number of columns and rows required − depending on the
required detail of each picture,
5.12
•   Set the screen mode for the resultant “pages”,
5.12
•   Choose the path/name format to name each picture.
5.12
You can drop directories onto the icon and several pages are generated.
The pages can be stepped through to view and saved to make a hardcopy of
each page. Alan Wilburn, Hartlepool.
5.12
• Auto destruct! − In the Hints and Tips section of the June edition of
Archive, there was a mention of possible problems with the auto-save
facility of Impression under the title “Auto-destruct!”. Version 2.16
(and 2.17) of Impression does not appear to be as dangerous in this
respect. If the auto-save facility is turned on but with the “With
prompt” switch left off, a prompt will still appear on-screen before the
first time the document is auto-saved.
5.12
This also occurs the first time the document is auto-saved after it has
been reloaded. It is mentioned in the Impression 2.15 release notes on
pages 1 and 3.
5.12
Still on the subject of Impression, I have two problems that readers of
Archive may be interested in:
5.12
Spelling checker − If you check and replace a misspelled word by using
<Ctrl-W> to invoke the spelling checker, this appears to work correctly
except that if you have typed nothing after the word, it also has the
effect of <Ctrl-B> and returns you to basestyle. Obviously the solution
is to always type at least a space after the word before using <Ctrl-W>.
5.12
PrintBJ (the extension module to print draft text) − This makes a
reasonable attempt at interpreting styles as far as different fonts and
font sizes are concerned but seems unable to cope with rulers or effects
that control the position of the text on the page. Tabs it gets
completely wrong, right aligned text is centred, centred text is centred
but using a page width about three quarters of the actual width. This is
using the Canon BJ-10e. I have phoned Computer Concepts who suggested I
wrote enclosing example documents both printed and on disc. This I did
several months ago and have not received a reply. Has anyone else
experienced these problems or indeed found a solution? Nick Edgar,
Doncaster.
5.12
I can understand Nick’s problems and only print out in graphics format.
My method of working for proofing documents is always to print at a
lower resolution − and hence achieve a quicker print out. Users of the
Deskjet 500 have the option of using the draft mode on the printer for
this purpose and leaving the resolution set on the driver unchanged.