Partition Manager

Unknown or missing variable in “Partitions.GPT”

What line was the error on?

The second FAT32 partition is not detected by PartMgr.

Could you create a debug log please and post or email me the result. Set DEBUG%=TRUE in !RunImage

160 Drive_Info%(D%, D_Partitions%)=0

Interesting error reporting from BASIC

If the Iyonix’s main disk was filecore formatted to say 128GB and then Fat32 formatted for the rest – would the Filecore 128GB run slow?

If the Iyonix’s main disk was filecore formatted to say 128GB and then Fat32 formatted for the rest – would the Filecore 128GB run slow?

Hmm, if the disc was “partitioned”¹ would the Iyonix know that there were two “discs” to look at on the one physical spinning rust?

¹ My understanding is that RO knows absolutely diddly about partitions and assumes the first bit of disc space is it. Could be wrong, but the filesystem bounty has making the system partition aware as part of the work.
Quote: “Building on the foundations laid in step 1, this next step will bring support for industry standard media partitioning to the native RISC OS filing systems.”

What line was the error on?

Sorry, I forget … 136.

Could you create a debug log please…

You find here

If the Iyonix’s main disk was filecore formatted to say 128GB and then Fat32 formatted for the rest – would the Filecore 128GB run slow?

No, the first 128GB is fast, the rest is slow. As long as the filecore partition is all within the first 128GB, it’s fast.

Second partition is FAT32 made with Antons partiFC

Unless I’m missing something, the debug logs don’t show any issues.

ADFS::4 is the only drive showing a FAT partition and is 500GB, is that the drive you’re referring to?
Partition 0 is Linux (type &83) &800-&00E09C800 (~120GB)
Partition 1 is FAT (type &B) &00E09D000-&02C2E9000 (~240GB). FileCore can’t access the last LBA, so it should be shown in red?

160 Drive_Info%(D%, D_Partitions%)=0

Interesting error reporting from BASIC

Are you saying that line errors? What’s the error?

Sorry, I forget … 136

Line 136 in GPT.Partitions errors?

Line 136 in GPT.Partitions errors?

I think …

The GPT drive is connected to SCSI.

ADFS::4 is the only drive showing a FAT partition and is 500GB, is that the drive you’re referring to?

No. ADFS::4 is detected correct. I ignore Linux in RISC OS and mount the second partition with fat32fs:mount -fp2 :12 { > null: }.
ADFS::7 is the drive I mean. Behind the FileCore partition is a FAT32 one.
I mount with fat32fs:mount -fp2 :15 { > null: }

I changed the order of the devices. Plug GPT as ADFS::7 and “ADFSBackup” to SCSI.
No error message.

GPT is unknown now. Is the same output if I use the GPT device on ARMX6 and try PartMgr.
…
Hope I have uploaded the right debug file.

Nice if BASIC gave the name of the variable missing.

Is the Disc formatter a RO Prog – and available for download?

Is the Disc formatter a RO Prog – and available for download?

I think you mean the FAT partition behind FileCore…
The tool is Anton Reisers partiFC. Is free but use at your own RISC and risk ;-)
The partiFC tool recognize a FileCore format and creates a MBR parition table with a protection partition for the FileCore Disk and a FAT32 partition for the otherwise unused disc space.
PartiFC you find here
I use it with the regulary FAT42FS and mount the new partition like the discription in the readme/help.

Nice if BASIC gave the name of the variable missing.

There is no variable where one would be expected, so no name available. ;-)

FAT 42 the answer to everything – Hitch hiker guide to the Galaxy :-)

Doesn’t BASIC know what Variable it’s looking for?

Nice if BASIC gave the name of the variable missing.

I’m still waiting for BASIC to report error line numbers when in the desktop (you’ll notice subtly different behaviour between that and command line use).

so no name available. ;-)

Yes, there is.
BASIC is looking for something. It fails to find it. Report what it failed to find. Might make things simpler when “Unknown or missing variable ‘partiton%’ at line 123”, one look and you can facepalm and add the missing ‘i’. ;-)

Nice if BASIC gave the name of the variable missing.

Seconded. Here my approach to deal with the deficiency, is to add a *report on the line before and list all the variables in the erroring line. Reporter has the advantage of NOT breaking when the variable doesn’t exist, so it makes it possible to see which one (or more) is undefined. Reporter does break on array(y) when y doesn’t exist (or I may misremember, it might be when array() doesn’t exist)

Fixing BASIC would be much better.

Yes, there is.

Oh, no there isn’t. ;-)

Try something simple such as WIDTH without a parameter. There is nothing to find there. It is “a missing variable”, or more strictly a missing expression.

I agree that the four-decade old error reporting is showing its age. Perhaps printing the line of the error and highlighting the program pointer would be good.

When programming with Reporter and throwback I do not have any problem with errors as they are.

I suspect the ‘normal’ error is where a variable name is mistyped, rather than omitted with WIDTH (or similar). Where BASIC has looked for a variable name but not found it, I think that when the error is detected registers hold the first character and the address of the rest of the string used. In those (common) cases it would seem possible to include the variable name in the message.

Another similar message I find infuriating is File not found

Perhaps we should add ‘but can’t be arsed to tell you what the filename was and I have erased my memory of its name now. Nerr!’ I can remember when the BASIC error messages were limited to two characters such as ‘FC error’ so we have it easy.

FAT 42 the answer to everything – Hitch hiker guide to the Galaxy :-)

Doesn’t BASIC know what Variable it’s looking for?

These are not the variables you are looking for :) (Star Wars the abused version)

I suspect the ‘normal’ error is where a variable name is mistyped, rather than omitted …

I suspect there may be more omitted errors than you think, certainly for me.

it would seem possible to include the variable name in the message.

I am sure it would, but first you would need two new errors for “Variable not found” and “Expression omitted”, which is the dual purpose of the current error. I suspect that the “Variable not found” part only arises from the EXPR sub-routine. The “Expression omitted” part is quite widespread.

0.66 now available. I’ve resolved the “missing variable in GPT.Partitions line 146” issue and also fixed a bug that could cause partitions created in gaps to overlap the next partition.

@Raik see if this resolves either the ADFS 7 partitions not being reported correctly, or the missing FAT after the ADFSBackup partition.

0.66 now available.

Thank you for adding the version number to the title bar – that will be a great help to us NetSurf downloaders!.

And my RPi 50MB boot “partition” now shows-up as well!

Sorry for this…

@Raik see if this resolves either the ADFS 7 partitions not being reported correctly, or the missing FAT after the ADFSBackup partition.

GPT nearly ok (partitions names are SSDp1 and SSDp2), ADFS 7 ok, but ADFS 5 + 6 are gone…

Debugfile

ADFS 7 ok, but ADFS 5 + 6 are gone

I wondered if fixing ADFS 7 might break something. It’s broken full disc FileCore drives formatted with recent versions of HForm. The problem is HForm writes a malformed MBR which has no partition entry for the FileCore area, so it’s indistinguishable from an unpartitioned MBR drive…hence ADFS 5 and 6 are now showing as unpartitioned.

I’ll modify it to detect an MBR with “Castle” in the OS field that has no partitions, and just assume it’s full disc FileCore. I also need to completely rewrite the FileCore detection to not rely on the OS at all, which is a slightly bigger piece of work. I’ll try to get the former fixed later today.

What did you use to partition SCSI 0:0? The GUID’s are the not the official FileCore value listed in PartMan’s source.

What did you use to partition SCSI 0:0? The GUID’s are the not the official FileCore value listed in PartMan’s source.

Is the “backup mirror” of my ARMX6 SSD.
The SSDs are formated with one of the MX6 tools around the partman stuff, I mean gpttest (I can not take a look at this time). Maybe there is a newer version of gpttest but I newer try. The drive works without problems since I try an earlier version.

0.67 now available. It checks for an HForm generated MBR with no partitions and falls back to FileCore full disc.

The SSDs are formatted with one of the MX6 tools around the partman stuff, I mean gpttest

The gpttest source uses the correct GUID. Did you self-compile it, or was it supplied with the MX6? I’m trying to understand if we have two “official” GUID’s for FileCore that I need to handle, or if yours is a one-off that I can ignore.