md5 checksums

I have 2 files decompressed from the same compressed file (extension gz), one decompressed on a Linux system and the other on RISC OS, but they have different md5 checksums. Wierd or what?

I found the RISC OS md5sum module inside !CoreUtils, but will newcomers to RISC OS know where to find it when they want to check downloads from this ROOL website against the md5 checksums published on this website? And, even if told that that it’s part of !CoreUtils, will they know where to find !CoreUtils?

It might depend on the unpacked contents, are things getting converted to RISC OS names/types? Then the original MD5 would have to be also generated on RISC OS. If it was a singular file like an disk image or iso, the checksum should be the same.
I was just thinking this morning that a checksum routine on installed modules and perhaps other parts of the machine
could be like a health check, handy on flash based machines,
or for checking the Iyonix before reflashing the roms.
A table of checksums would have to be maintained for versions
of modules or roms.

The original is a compressed disc image and so after decompressing there is the disc image which is a single file. So the only file type involved is that of an image file, i.e., .img . I ‘burned’ the one decompressed using Linux to an SD card and it works. Haven’t tried the other decompressed file.

I have 2 files decompressed from the same compressed file (extension gz), one decompressed on a Linux system and the other on RISC OS, but they have different md5 checksums. Wierd or what?

Why not compare the two files using something like ‘fc’ on windows to see whether they are of identical length and contents?

David, Have you tried the alternatve (to Spark) gz?
http://www.riscos.info/
There is also a port of md5sum for RISC OS which would rule out the possibility of an error while copying the file back to Linux, assuming that is where you are comparing the checksums, but I’d try the different gz and the same md5 checker initially.

I have done some testing using gzip to compress and decompress a small text file and the (large) disc image file on both RISC OS and Linux (EeePC running Xandros) calculating the md5 checksums in each case. The check sums agree on both paltforms. Perhaps the original discrepancy in the md5 checksums was due to decompressing the disc image file with .gz extension using !SparkFS instead of gzip. Could this be the cause of the error “Duplicate Character in translation table” when !SparkFS decompressed the file? (see “Duplicate Character in Translation Table” thread in Community Support). My Pandaboard can now run Ubuntu 12.04, at least it boots up – haven’t had time to test anything in the ‘distro’.