Replay: BIOS questions

Question from Ibn Numberaz: Is it possible to work in files in the BIOS that reach a length of more than 16MB? The length of the file descriptor depends on this
by as much as 2 bytes ;-) What do you mean: how many bytes should be allocated to a
position (and length) in the file: 3 or 4? Still, 2 bytes (64Kb) is not enough,
but will anyone on Specky need files longer than 16MB?

With respect, Roman aka Megus of Brainwave Association.

[Team ZX-Spectrum] [Team Agatha Christie] [Team Metallica] [Team Iron Maiden]

19 Aug 98 Roman Petrov, hoping that Everyone would read, wrote:

RP> and for as many as 2 bytes ;-) You mean how many bytes
RP> should be allocated to the
RP> position (and length) in the file: 3 or 4? Still, 2 bytes
RP> (64Kb) is not enough,
RP> but will anyone on Specky need files longer than
RP> 16MB?

No, there’s no need to pose the question like that imho. It’s not a question of
“will someone need...”. I would like that in the future NeOS
doesn’t run into such phenomena as the limitations of descriptors.
Look like is-dos ran into problems with the size of the screw. But excesses, it seems, are also useless; everything for Speck should be compact. What is the total length of the descriptor? - is it worth saving two bytes?

All the best.
Roman.

At 11:40, Thursday August 20, 1998 there was such a thing that
Roman Milukov wrote a letter to Roman Petrov:

RM> No, there is no need to pose the question like that imho. It's not a matter of
RM> “will someone need...”. I would like that in the future NeOS
RM> does not run into such phenomena as the limitations of descriptors.
RM> Look, how is-dos ran into problems with the size of the screw. But excesses,
What was their emphasis? I haven't worked seriously with iS-DOS, so I don't know much. I don’t think there should be such a limitation - you can
support up to 65536 clusters (this is a full 16-bit FAT on MS-DOS). Cluster
in TR-DOS - 1 sector, still enough. I don’t know about iS-DOS :(
RM> it’s also useless, everything for Spec should be compact. And what
RM> is the total length of the descriptor? - is it worth saving two bytes?
36 bytes with support for up to 4Gb, 34 bytes with support for 16Mb. Probably all
worth leaving on 4Gb.

With respect, Roman aka Megus of Brainwave Tracks.

[Team ZX-Spectrum] [Team Iron Maiden] [Team Die Krupps] [Team Guitar]

RP> What was their focus? I haven’t worked with iS-DOS

I don’t know for sure, but it seems that the size of a block device there is limited to 16M.
Therefore, the screw there has to be divided into several devices of <16M each. The maximum
is possible, it seems, 6 devices (minus the flop and ramdisk).

...
RP> should - it is possible
RP> to support up to 65536 clusters (this is a full 16-bit FAT
RP> on MS-DOS). Cluster

Here's the complete one you need!

All the best.
Roman.

├┤i Roman & Roman !

[...]

RP> What was their emphasis? I haven't worked seriously with iS-DOS, so
RP> I don't know much. I don’t think there should be such a limitation - you can
RP> support up to 65536 clusters (this is a full 16-bit FAT on MS-DOS).
Cluster
RP> in TR-DOS - 1 sector, that’s still enough. I don’t know about iS-DOS :(

Your device size is limited to 16MB -->>> make the cluster
larger (if necessary).

The number of 'block devices' in the isdos is limited to 8 pieces.
16(MB)*8(devices)=128 MB screw size. :-(

I hope this won’t happen? That’s why we need larger clusters
and devices from A to Z. 26*16=416MB - already better.

RP> 36 bytes with support for up to 4Gb, 34 bytes with support for 16Mb. Probably still
RP> should be left at 4Gb.

It is very necessary!

W!7h Be$7 W!$he$, FK0.

At 01:24, Saturday August 29, 1998 there was such a thing that
Kirill Frolov dashed off a letter from Roman Petrov:

KF> Your device size is limited to 16MB -->>> make the cluster
KF> larger (if necessary).
Wait, nothing is limited. For TR-DOS it is enough, for iS-DOS it is the same.
And MS-DOS has clusters of different lengths. After all, on the PC 65536
they make do with clusters - so we will make do, there will also be large clusters.

KF> I hope this won’t happen? Therefore, we need large clusters
KF> and devices from A to Z. 26 * 16 = 416 MB is already better.
Well, well ;-) Devices from A to Z, and add megabytes - as much as your heart desires. For example, on my PC there are 4 3.2Gb drivers. It will be possible to do the same
on the Spectrum.

Thu Aug 27 1998 16:49 you wrote to Roman Petrov:

RM> I don’t know for sure, but, it seems, there the size of a block device is limited to 16M.
RM> Therefore, there the screw has to be divided into several devices of <16M each.
RM> The maximum possible, it seems, is 6 devices (minus flop and ramdisk).

8 minus. If you have a hard drive, the ramdisk is no longer so vitally important; disk drives can also be reduced to 1-2. it turns out, roughly speaking, a hundred megabytes.
I wonder what something so exotic this might not be enough for, especially considering
that you can quickly attach any other partition to any letter.

RP>> it should - you can support up to 65536 clusters (this is a full 16-bit
RP>> FAT on MS-DOS). Cluster

RM> That's the complete one you need!

65535*2=128k - it turns out that you just read fat and have already run out of memory? ;)

Bye, Michael.

RM> ramdisk).

MK> 8 minus. Ramdisk in the presence of Windows is no longer so much

You know better. ;)

MK> is vital,
MK> drives can also be reduced to 1-2. it turns out roughly
MK> saying a hundred megabytes.
MK> I wonder what exotic things this might not be enough for
MK>, especially considering

What does “not have enough” have to do with it? I think, personally, 40M would be enough for me.
But I was stumbling here... A good screw for 40M costs as much as a
good screw for 210M. :) Yes, and now try to find less than 210. And do you think I
will be pleased to realize that “roughly speaking” only half of the screw
is available to me?!

And since you’re here...
I asked you with soap, but you obviously don’t read it.
Tell me, why didn’t you do ZX_ZIP/UNZIP under is-dos?
(And a PC-compatible ZIP? Did Anzip make it?)
But somehow everything turns out inconveniently: in order to send is-dos files, you need to throw them into
TR-DOS, zip them up, and then back to is-dos, since your new soaper is there...

All the best.
Roman.

At 00:06, Wednesday August 26, 1998 there was such a thing that
Denis Dmitriev wrote a letter to Roman Petrov:

DD> How many people will have time to “thank” you for such a limitation,
DD> before a new version of the subject is released? :) (and the new version of _already_
DD> will be incompatible with the old one, because the file length field will increase, and
DD> all the fields behind it will shift...) So think ahead!
So you've already decided - 4Gb

Fri Sep 04 1998 11:05 you wrote to Michael Kondratyev:

RM> What does “not have enough” have to do with it? I think, personally, 40M would be enough for me.
RM> But I was stumbling here... A good screw for 40M costs the same as
RM> for a good one for 210M. :) Yes, and now try to find one less than 210. And
RM> do you think I will be pleased to realize that “roughly speaking” only half of the
RM> screw is available to me?!

I’ll cut to the bottom - _at the same time_ ~100m are available. There are as many _total_ available as there are
partitions created. Changing a partition on any letter of the device is a time-consuming operation that is hardly more troublesome than, for example, changing a floppy disk in a drive. for
scorpios there is a limit of 63 sections - with them it will be about 1g. but I can equally well express my opinion if I find a hard disk drive, let’s say 2.1g

RM> And since you’re here...
RM> I asked you with soap, but apparently you don’t read it.

some amount was lost, but in general I read it ;)

RM> Tell me, why didn’t you do ZX_ZIP/UNZIP under is-dos?

loss of meaning. it is a priori specific and intended for trdos, and therefore I do not
see the benefit of such an act.

RM> (And a PC-compatible ZIP? Did Anzip make it?)

memory. if you do it, then the correct method, and for it you are welcome somewhere like
at least 150k. In principle, I’ll try to watch it again sometime; if it fits
inScorpio free, that is, a chance of appearing.

RM> Otherwise, everything turns out somehow inconveniently: send is-dos files so that
RM> you need to throw them into TR-DOS, zip them, and then back to is-dos, once
RM> your new soaper is there...

as a quick, very temporary plug for this grief, you can make a machine for making zips without compression. There is at least one plus - it will not be .pkt (which means
there will be the possibility of recovery in case of communication failures). besides, v42bis
compresses not nearly as bad..

Bye, Michael.

MK> memory. if you do it, then the correct method, and for it
MK> you are welcome somewhere like
MK> at least 150k. in principle, I’ll try to watch MK> again sometime; if
MK> fits into Scorpio's free time, then there is a chance of an appearance.

And you do it under Pent512K, Profi... Right now, and KAYI with 1M seem to be doing...

MK> as a quick, very temporary plug for this grief, you can
MK> make a manufacturing machine
MK> zips without compression. there is at least one plus - it's not

It's not interesting at all without compression... :)

MK> besides, v42bis
MK> does not compress so badly..

But even ZXZIP does it better.

All the best.
Roman.

~~~~~~~~~~~~~~~~~~~~~~~~
this is not good. Why not make some kind of stripped-down
version that needs fewer ?

MK> zips without compression. there is at least one plus - it will not be .pkt (and
meaning there will be
MK> the possibility of recovery in case of communication failures). besides, v42bis
is not nearly
MK> as bad at compressing..

Well, yes... On a PC, mail is also well received and readable,
maybe even better than on the Spectrum...

W!7h Be$7 W!$he$, FK0.

RM>
RM> But even ZXZIP does it better.

v42bis really presses well, I used 2400 mods
unzipped with compression, and zipped (pc zip) without
compression - I didn’t notice much of a difference.

See you. Jan Zagorskikh.

Wed Sep 09 1998 01:00 you wrote to Michael Kondratyev:

KF> this is not good. Why not make some kind of stripped-down
KF> version that needs less?

yes, you can, for example, throw out the hashtables and let them search for three years.. it will be even
an order of magnitude slower than, say, doing a simple move one bit at a time.

Bye, Michael.

RM> You shouldn’t be doing this.
RM> He serves very faithfully in TR-DOS! And in is-dos it would be useful, in the absence of a
RM> PC-compatible.

there you need and can be useful either again specific (which you don’t want because, after all
the disk system is much closer to the common ones), or really
compatible.

RM> came in handy, in the absence of
RM> PC-compatible.

MK> there is needed and can be useful either again specific
MK> (which you don’t want because after all
MK> the disk system is much closer to the common ones), or
MK> really compatible.

Of course, I don’t want to be specific - again not like everyone...

Tue Sep 15 1998 12:38 you wrote to Michael Kondratyev:

RM> So the issue is resolved? Will people have PC-compatible spare parts? ;)

the issue can be resolved, but only after the fact. the start of work is not yet planned.

RM> Yes, I just don’t want to rip an uncompressed file. And it’s not always convenient to take away space on the screw from the boss
RM>.

with yue - maybe so. Archmail, you understand, is still immediately unpacked and
deleted.

Bye, Michael.