Page 1 of 1

UBCD V5.2a3 available for download

Posted: Thu Dec 27, 2012 5:42 pm
by Victor Chew

Re: UBCD V5.2a3 available for download

Posted: Thu Dec 27, 2012 11:43 pm
by Explorer09
Consider upgrading these in the next UBCD alpha release:
  • HDAT2 4.9.3 (released this Christmas, i.e. 2012-12-25)
  • Parted Magic 2012_12_25
Also, I see some errors in the licenses of some software in this page http://www.ultimatebootcd.com/v52/index.html
  • Mersenne Prime Test : It's not really Public Domain. It's under a permissive free license. I think it's better to write "free license", "free software" or "permissive free" or something like that in the license field.
  • KeyDisk: It's freeware, not GPL.
  • PC CMOS Cleaner: It's freeware, not GPL.
  • PCLoginNow: It's freeware, not GPL.
  • PCRegEdit: It's freeware, not GPL.
  • PC Disk Eraser: It's freeware, not GPL.

Re: UBCD V5.2a3 available for download

Posted: Fri Dec 28, 2012 7:56 pm
by Victor Chew
@Explorer09: Noted. Thanks!

Re: UBCD V5.2a3 available for download

Posted: Tue Jan 01, 2013 1:23 am
by pica
It looks like from my error message that you updated ISOLINUX to 4.06

>2012-10-23 : Syslinux 4.06 released. This was a bug fix release, but some new features were added.

(I'm having issues putting the alpha on a USB flash, but I think that's because my hashes don't match. Two different d/l and two different hashes. Maybe the third time will work?)

Re: UBCD V5.2a3 available for download

Posted: Tue Jan 01, 2013 3:10 am
by Victor Chew
Note: I am posting this PM from ady (and another one after this) because PMs are not useful for tracing decisions later.
ady wrote:Victor,

I asked a similar question in the public forum before your released of ubcd52a3.iso. Since there was no answer, I'll ask here again.

I can improve the fs format of the new superfloppy image. I have identified already at least one parameter that can be improved without any negative effect at all, but for others I need your general guidance.

For example, in ubcd52a3, the superfloppy is Cx2x36. Have you specifically / intentionally selected such geometry (maybe for some better boot or BIOS compatibility)? Or is it just because of some limitation of the tool you used? (FWIW, "real" superfloppies use other geometries, but maybe you selected it because you might know of some better compatibility by using that specific values?)

Another example of your selected values is FAT12. While it is not strictly necessary, would you consider FAT16 to be less "compatible" for BIOS? I can play between cluster size, number of clusters and other parameters, and in a certain situation, FAT16 might be a valid preferable option. But if you tell me that FAT12 might improve the chances of bootability, then I would take that into consideration.

There are more parameters of the superfloppy I could improve, if I just knew about specific conditions needed for better bootability.

You might have used some values for your superfloppy just for the possible limitations of the tool you used. If that's the case, I can certainly improve the fs format.

So, if you are so kind to let me know the conditions that are _really_ needed (besides the available fs size, which is about 14MB), I'll be happy to prepare an alternative superfloppy (or several if you want).

Best Regards,
Ady.

Re: UBCD V5.2a3 available for download

Posted: Tue Jan 01, 2013 3:11 am
by Victor Chew
And the next...
ady wrote:Hi Victor,

I have uploaded an alternative superfloppy image.

Differences with your previous superfloppy geometry:

Your superfloppy (in UBCD52a3):
Sectors_Per_Cluster: 32
Sectors_Per_Track: 36
Heads_Amount: 2 (0-1)
Cylinders_Amount: 512 (0-511) <== (too big for floppies in MSDOS)
Total_Sectors_Amount: 36864
Sectors_Per_FAT: 4
Amount_of_FATs: 2
Root_Directory_Sectors: 17
Max_Root_Directory_Entries: 272
Clusters_in_data_area: 1151

New superfloppy (linked to download, keep reading):
Sectors_Per_Cluster: 16
Sectors_Per_Track: 63
Heads_Amount: 16 (0-15)
Cylinders_Amount: 32 (0-31)
Total_Sectors_Amount: 32256
Sectors_Per_FAT: 6
Amount_of_FATs: 1
Root_Directory_Sectors: 9
Max_Root_Directory_Entries: 144
Clusters_in_data_area: 2015

The new image is smaller and compatible with ATA and BIOS geometry standards. It should be (unnoticeable) more efficient (only one FAT copy) and it wastes less space (only one FAT copy, less unnecessary root entries, and smaller cluster size).

The amount of Root Directory Sectors currently really necessary is 1 (one), and one only. The reason to have more than one Root Sector is to complete the last cylinder of the image. Using complete cylinders helps reduce potential incompatibilities.

The new image was already tested (by me) to support the current requirements for size and amount of files, and its geometry is correctly identified by memdisk and by grub4dos.

I would also suggest to consider the potential “pros” and “cons” of changing the APPEND line in the syslinux menu for the memdisk fdubcd boot entry,
from the current:
APPEND raw
to:
APPEND raw floppy
or to:
APPEND raw floppy c=32 h=16 s=63
or even to:
APPEND raw floppy c=32 h=16 s=63 ro

I shall be clear; I am not convinced that adding all those parameters is really necessary. It may help in some case, but if the geometry is changed then the entry should be corrected accordingly (or if someone wants to use the image uncompressed and not read-only). In most systems, the additional APPEND parameters should not be necessary. The only reason I would consider them is for some (whether too old or too new) buggy BIOS, so probably the best compromise would be something like the following lines (one effective line, and two lines commented out):
APPEND raw floppy
# If the above fails, try adding some additional parameters
# APPEND raw floppy c=32 h=16 s=63 ro

In any case, this new image should be tested in as many systems as possible, from CD and from UFD.


Additionally, in my image:
Volume Label: FDUBCD_147 (The underscore is for aesthetic reasons for utilities that can display the volume label as file name and extension, "FDUBCD_1.47", instead of “FDUBCD14.7”.)
Boot code source: sys.com version 3.8pre2012MAY11
OEM String: FRDOS5.1 (as used in latest format.com and sys.com 3.8pre).
Serial Volume: all zeroes (It is a RAM read-only superfloppy. There is no utility to other serial volume values, and using all zeroes helps its compression).
Empty data clusters are filled with “F6” for better compatibility with floppy tools.


The initial (non-bootable) geometry of the FAT12 image can be achieved by several methods, including mtools version 4.0.17 (either under Linux, or using the Windows port from Icecube).

In this opportunity, I happened to use different tools, not mtools, but mtools is capable of building this exact CHS geometry and the FAT12 fs.

It should be noted that there are other alternative geometries that can successfully serve as superfloppy image for fdubcd; each with its own advantages and disadvantages.


The file name of the uncompressed image I have uploaded is already fdubcd.img. To avoid accidental overwriting, I changed the name of the compressed gzipped image, from "fdubcd.img.gz", to "FDUBCD_147-32256.img.gz". So after downloading the gzipped image, you can just change its name back to "fdubcd.img.gz" and build a new UBCD test for you.


The file will be available for the next 7 days at:

DOWNLOAD:
http://www.fileconvoy.com/dfl.php?id=g8 ... 82970ac679

SIZE: 14'187'395 bytes (13.53 MiB)

MD5: e82daaf272f084e1ae3795153ae13e37



If you have any questions, please don't hesitate to ask.

I'll be posting any feedback / reports I may find in UBCD 52a3 in the public forum, specially if I find something about fdubcd.

Best Regards,
Ady

EDIT / UPDATE:
I just tested the build of mtools 4.0.17 that Icecube made for Windows users, and for replicating my image (in fact, generating one image anew), a FAT12 image with the suggested geometry, it has some problems in the resulting image. I don't know if the Linux version has / generates those problems too.
On one side, it is a good thing that I used a different method for my image (the one I uploaded), because the result is a nice, clean, correct, valid, bootable FAT12 superfloppy image. On the other side, providing a simple mtools command is not enough to replicate this geometry accurately enough, so now I have to think of a simple step by step method for you to replicate it, unless you are able to do it by yourself with your own tools.

Re: UBCD V5.2a3 available for download

Posted: Tue Jan 01, 2013 3:23 am
by Victor Chew
ady wrote:I would also suggest to consider the potential “pros” and “cons” of changing the APPEND line in the syslinux menu for the memdisk fdubcd boot entry,
from the current:
APPEND raw
to:
APPEND raw floppy
or to:
APPEND raw floppy c=32 h=16 s=63
or even to:
APPEND raw floppy c=32 h=16 s=63 ro
As to this point, I am inclined to leave the current parameters unchanged until some concrete error reports come in subsequently to convince us otherwise.

Re: UBCD V5.2a3 available for download

Posted: Tue Jan 01, 2013 5:46 am
by pica
Third try wasn't the charm. I used curl on the command line so I could figure out the issue. It exited with a code 18 (Partial file) error. There's something wrong with that mirror.

I was able to restart the download where the mirror dropped me by issuing

Code: Select all

curl -C - "http://ubcd.mirror.fusa.be/ubcd52a3.iso" -o "./ubcd52a3.iso"
Both hashes match now. If you are downloading double check the hash.

Re: UBCD V5.2a3 available for download

Posted: Thu Jan 03, 2013 5:43 pm
by JonaLinux
Linux Freedom has made UBCD V5.2a3 available for download at http://linuxfreedom.com/ubcd/download.html

Re: UBCD V5.2a3 available for download

Posted: Sat Jan 05, 2013 3:53 am
by ady
Some comments / suggestions:

1_ Change fdubcd.img and fdubcd.img.gz to fdubcd.ima and fdubcd.ima.gz. Since the image is not 1440 nor 2880 KiB, the use of "ima" as extension should give users a hint about being a floppy image (not a HDD image, as usually expected from the “img” extension). It may also help reduce the chances of opening the image with an inadequate program by the file extension association.

2_ When "ima" is used, the scripts and menus should be also changed accordingly.

3_ Should ETtool, ETDump (and maybe FindCD) be added to the NwDsk menu? Or maybe they are already there and I don't know it? Or maybe they are even not worth to be included at all in fdubcd if not using fdubcd.iso?

4_ fdubcd.img.gz -> fdubcd.img -> DOSAPPS -> UBCD.INI -> "Version = 5.2a2" was also used in 5.2a3.

5_ Since the fdubcd version was upgraded from v1.44 in UBCD 5.1.1 to v1.47 in UBCD 5.2a3, are there any changes or updates other than the ones related to manage the different types of fdubcd images (ISO+floppy - HDD - superfloppy)?

6_ Adding a new syslinux comment line:

Code: Select all

# APPEND floppy raw c=32 h=16 s=63 ro
after the main boot entry for FDUBCD, should not affect anyone and it could be useful for troubleshooting. The other possibility I could think of, instead of the above, is adding:

Code: Select all

# please report any issue with this fdubcd image at http://www.ultinmatebootcd.com 
or to the forum url directly, without any extra tip, so _maybe_ problems would be reported :?: . In any case, I would certainly use at least "APPEND floppy raw " (including "floppy") in that APPEND line for MEMDISK..

Re: UBCD V5.2a3 available for download

Posted: Sun Jan 06, 2013 12:14 am
by ady
ady wrote:3_ Should ETtool, ETDump (and maybe FindCD) be added to the NwDsk menu? Or maybe they are already there and I don't know it? Or maybe they are even not worth to be included at all in fdubcd if not using fdubcd.iso?
Independently of the above comment, there is a duplicated ETTOOL in 5.2a3. It can already be found inside UTILS.CAB, so, is the one in fdubcd -> BIN really needed? What about ETDUMP (and FindCD)?

*****

It might be worth updating fdubcd -> BIN -> 7ZA.EXE to 9.20.1 from

Code: Select all

http://ftp.ibiblio.org/pub/micro/pc-stuff/freedos/files/util/file/7zip/9.20.1/testing/p7z9201-latest.zip
Changelog since p7zip version 9.13 (currently used in fdubcd v1.44 in UBCD 5.1.1):

Code: Select all

http://sourceforge.net/projects/p7zip/files/p7zip/9.20.1/
(The README is displayed directly down in that page)
Just as a remainder, this is the version for DOS to be used in fdubcd; not to be confused with the one in the auxiliary tools of UBCD.

Then in fdubcd -> AUTOEXEC.BAT, line #94, I would suggest changing from:

Code: Select all

if exist %srcdrv%\bin\7z.exe goto _extracok
to

Code: Select all

if exist %srcdrv%\bin\7z*.exe goto _extracok
(Note the added "*" so to cover several possible variations like 7za, 7zr, 7zdec,... This code was never corrected before because the condition is already triggered by "extract.exe", but since we are at it, there is no reason not to do it. Please add either "*" or at least "a".)

Then in the same AUTOEXEC.BAT, line #199, I would suggest changing from:

Code: Select all

echo AUTOEXEC: Extractor 7-ZIP 4.57 (.7z) by Igor Pavlov
to

Code: Select all

echo AUTOEXEC: Extractor 7-ZIP 9.20.1 (.7z) by Igor Pavlov and p7zip
(Note the version and also the addition of "p7zip", so to make it easier to track down the source for future updates.)

Re: UBCD V5.2a3 available for download

Posted: Tue Jan 08, 2013 9:55 pm
by Victor Chew
ady wrote:Change fdubcd.img and fdubcd.img.gz to fdubcd.ima and fdubcd.ima.gz
Don't think it's necessary. There are plenty of floppy images in the "images" subdir with extension ".img.gz" (eg. hdat2.img.gz, plpbt.img.gz), so it should be obvious that's the current convention.
ady wrote:Should ETtool, ETDump (and maybe FindCD) be added to the NwDsk menu?
I am actually going to remove them in the next release, since they were there for debugging the eltorito.sys problem in the first place.
ady wrote:fdubcd.img.gz -> fdubcd.img -> DOSAPPS -> UBCD.INI -> "Version = 5.2a2" was also used in 5.2a3.
Thanks! Will be fixed in next release.
ady wrote:Since the fdubcd version was upgraded from v1.44 in UBCD 5.1.1 to v1.47 in UBCD 5.2a3, are there any changes or updates other than the ones related to manage the different types of fdubcd images (ISO+floppy - HDD - superfloppy)?
Collated from previous changelogs:

5.2a1: Updated UMBPCI in FDUBCD to V3.83. Updated FDUBCD to use HimemX and JEMM386. Updated DOSLFN in FDUBCD to V0.41c.
5.2a2: Updated UBCD FreeDOS to V1.46. Updated KERNEL.SYS to Build 2041. dded support for "kblang" parameter in UBCD FreeDOS at Don Manuel's request for the OmniBoot project. \level1\keybrd.cab\bin\Keybrd.bat was modified for this purpose. Due to problems mounting FDUBCD ISO under grub4dos, this version introduces booting FDUBCD as a HDD image to work around this problem.
5.2a3: Updated UBCD FreeDOS to V1.47. This version tries booting FDUBCD as a 18MB superfloppy image created using the "FAT12build" script.
ady wrote:# APPEND floppy raw c=32 h=16 s=63 ro
OK. but without the "ro", since I think certain scripts want to write settings to the virtual floppy.
ady wrote:It might be worth updating fdubcd -> BIN -> 7ZA.EXE to 9.20.1
Done.
ady wrote:Then in fdubcd -> AUTOEXEC.BAT, line #94, I would suggest changing from:
All done!

Re: UBCD V5.2a3 available for download

Posted: Tue Jan 08, 2013 11:41 pm
by ady
Victor Chew wrote:
ady wrote:Change fdubcd.img and fdubcd.img.gz to fdubcd.ima and fdubcd.ima.gz
Don't think it's necessary. There are plenty of floppy images in the "images" subdir with extension ".img.gz" (eg. hdat2.img.gz, plpbt.img.gz), so it should be obvious that's the current convention.
There are some differences between fdubcd and the others. For example, when talking about "*.img" images of 1440 or 2880KiB, in most cases they will be assumed to be _floppy_ images - just as memdisk does. But the new fdubcd image is "much" bigger, so the assumption (by some users and/or some programs) might probably be that it is a HDD image - just as the documentation for memdisk states it (although I don't know if the actual behavior of memdisk acts as the documentation describes it).

In addition, it would be unusual for someone to edit or customize the other images without noticing their respective sizes (which already match floppy sizes). My guess is that the likelihood of some user wanting to edit or customize fdubcd would still be low, but not as low as other images in UBCD.
ady wrote:# APPEND floppy raw c=32 h=16 s=63 ro
OK. but without the "ro", since I think certain scripts want to write settings to the virtual floppy.
Fair enough. IMO, this particular entry certainly needs more testing / testers.

Re: UBCD V5.2a3 available for download

Posted: Wed Jan 09, 2013 2:49 am
by ady
The instructions for AleGr memory tester were
TEXT HELP
RAM tester able to test up to 3GB under DOS. When FreeDOS boots, select one of
the "Defensive" boot options. Then pass in the "/ignorexmm" parameter to
memtest.exe. For example, to test 3GB memory, type "memtest 2900 1 /fastdetect
/nousblegacy /ignorexmm" ("1" ignores first MB used by BIOS).
ENDTEXT
So, if the user follows the instructions, why is it that there is _one_ report stating that it can't work from fdubcd? I mean, isn't it possible that the user that reported a problem with this program is just not following those instructions?

Re: UBCD V5.2a3 available for download

Posted: Sun Jan 13, 2013 4:26 am
by Victor Chew
There are some differences between fdubcd and the others. For example, when talking about "*.img" images of 1440 or 2880KiB, in most cases they will be assumed to be _floppy_ images - just as memdisk does. But the new fdubcd image is "much" bigger, so the assumption (by some users and/or some programs) might probably be that it is a HDD image - just as the documentation for memdisk states it (although I don't know if the actual behavior of memdisk acts as the documentation describes it).
On the contrary, I think using a different extension will make the user think it is a non-floppy image. Using the same extension will make for consistency.
So, if the user follows the instructions, why is it that there is _one_ report stating that it can't work from fdubcd? I mean, isn't it possible that the user that reported a problem with this program is just not following those instructions?
You are right! I tested it using the "/ignorexmm" argument and it works fine. Thanks for pointing this out. Will re-add it to FDUBCD in next release.