Small issues with archive integrity.

After finding out another item removed because of DMCA I've decided to make a private copy of WinworldPC collection.

Took some time (few months), but it's finished now and as I have found out winworldpc.com is surprisingly small (at least files that you can download): slightly more than 8000 files, around 550GB total size. I kinda always assumed it would be much bigger.

For the future reference I have created ls-lR, md5, sha1 and sha512 files (attached).

Most files have verified fine, but there are few where checksums don't make any sense or don't match the content.

Download f8d2d0a4-031d-11ee-8470-0200008a0da4 “Textra 3.1 (1985) (5.25-360k)” has checksum from different download: 9753c54c-031e-11ee-8470-0200008a0da4 “Softerm Classic for DOS (1993) (3.5-1.44mb)”

Downloads 864df503-855b-11e9-ab10-fa163e9022f0 and d6bba27c-8556-11e9-ab10-fa163e9022f0 (both named “Palantir WinText and WinTime Demo (6-13-1988) (5.25-1.2mb)”) have the same checksum and send you to the exact same file on the download server… but only d6bba27c download actually succeeds! Probably limitation of a download system?

Download 6216a8f0-fd77-11ed-aca0-0200008a0da4 “myHouse 1.11 for DOS (1993) (5.25-1.2mb)” couldn't be downloaded at all (but has links!) while downloads 76cd168c-5f06-11eb-b764-0200008a0da4 “OpenVMS 8.4 Hobbyist Kit” and e35dac03-5f02-11eb-b764-0200008a0da4 “OS/400 V4R4” don't have any download links (yet, somehow, have a checksums? weird).

Really strange (and somewhat funny) situation is with download 41c38b75-c3aa-7d0f-11c3-a7c29d255254 “Borland Delphi 3.0 Standard (ISO)”: for some crazy reason this download has incorrect checksum… yet I have file downloaded years ago which have the exact same checksum as specified on the web site! Both files are valid archives, with the exact same size (254,743,859 bytes precisely!) and content… yet packaged form is wildly different… what kind of magic is that? Someone unpacked file and then packed it back? Why?

The other downloads where checksums don't match are:

c3a653c2-b039-78c2-a011-c3a7c29d2552 “Adobe Photoshop 5.0 (ISO).7z”
c872298c-8d4e-11ea-8c4a-fa163e9022f0 “AppleWorks 2.0 for Apple II Manuals (1986).7z”
299b6ed9-5f77-11eb-b764-0200008a0da4 “Bank Street Writer Plus 1.x Manual (1986).7z”
a59351e7-3272-11e9-8581-fa163e9022f0 “Claris MacWrite 5.0 (Jan 1988) (3.5-800k).7z”
0a15554b-9805-11ec-84e0-0200008a0da4 “Dac-Easy Accounting 2.0 for DOS (1987) (5.25-360k).7z”
ae0d0d72-613a-11ea-8c4a-fa163e9022f0 “Dr Solomons Anti-Virus Toolkit 7.74 for OS2 (3.5-1.44mb).7z”
cce5d681-18a2-11f0-bd4d-0200008a0da4 “FinalWord II 2.x Manual.7z”
95307287-1bf9-11e9-9b71-fa163e9022f0 “Hurricane 98 (1997).7z”
0466eefe-3aae-11ec-be67-0200008a0da4 “IBM Electric Poet 1.00 (1984) (5.25-180k).7z”
5eaa7d30-557c-11ec-a893-0200008a0da4 “IBM Writing Assistant 2.00 (1987) [French] (5.25-360k) (3.5-720k) (SCP).7z”
4971c0e6-63da-11eb-b764-0200008a0da4 “Lotus 1-2-3 9.0 Millennium Edition for Windows (1998) (3.5-1.44mb).7z”
ebe5b5c0-b5f4-11e9-b7f9-fa163e9022f0 “MathType 1.1a (1991) (5.25-1.2mb).7z”
15ab7e95-d740-11ea-8b14-fa163e9022f0 “Microsoft Commerce Server 2000 (Develop and Test) (2000) (ISO).7z”
1f5c777b-4c4a-11c3-a4c2-8d587054c392 “Microsoft Excel 3.0 (3.5).7z”
f3c99470-6ce1-11eb-b764-0200008a0da4 “Microsoft Excel 4.0 for Windows [Trad Chinese] (5.25-1.2mb).7z”
c04e8a15-3290-11eb-a665-0200008a0da4 “Microsoft MS-DOS 5.00 [BTI OEM] (1991) (3.5-720k).7z”
396fe0a0-ee9d-11ec-8dc3-0200008a0da4 “Microsoft Multiplan 4.01a (1989) (3.5-720k).7z”
bdd6a52c-c3c2-11ee-b155-0200008a0da4 “Microsoft Professional Tooklit for Visual Basic 1.0 (1992) (5.25-1.2mb) (3.5-1.44mb).7z”
483dc393-4225-c389-11c3-a6e280947e52 “Microsoft Windows 98 First Edition - Boot Disk (3.5-1.44mb).7z”
a837e2ba-8b1e-11e9-ab10-fa163e9022f0 “Microsoft Windows CE Toolkit for Visual Basic 5.0 and Visual CPP 5.0 (Mar 1998) (MSDN) (ISO).7z”
c9d9a40d-8b1f-11e9-ab10-fa163e9022f0 “Microsoft Windows CE Toolkit for Visual CPP 6.0 (ISO).7z”
2ab161cd-1201-11ea-9911-fa163e9022f0 “My Advanced MailList and AddressBook for Windows (1997) (3.5-1.44mb).7z”
8db090c8-b1d7-11ea-8b3c-fa163e9022f0 “Norton Utilities 7.0 (5-12-1993) [German] (3.5-1.44mb).7z”
cbedf74f-e3c7-11ee-b155-0200008a0da4 “Official Guide to Netscape Navigator (1995) [English] (ISO).7z”
007bac4f-7685-11ec-bb7d-0200008a0da4 “Oracle Developer 2000 for Windows (Forms 4.5) (1997) (ISO).7z”
aa023a89-2079-11ef-be8f-0200008a0da4 “Overhead Express 1.11R (1985) (5.25-360k) (Kryoflux) (SCP) (TC).7z”
b0225918-1207-11e9-9b71-fa163e9022f0 “pcANYWHERE 32 8.0 (ISO).7z”
7f34ac1b-a5aa-11e9-b7f9-fa163e9022f0 “PFS First Choice Document Conversion and Rescue Disk 3.00 (1989) (5.25-360k).7z”
fa3d0df0-a8f6-11ea-8b3c-fa163e9022f0 “PFS First Publisher 2.1 (1989) (3.5-720k).7z”
e01b36d1-c701-11eb-9de1-0200008a0da4 “RightWriter 3.1 for Macintosh (1990) (3.5-800k).7z”
626a12db-5b8c-11ec-a893-0200008a0da4 “Venix-86 2.0R (1984) [DEC Rainbow] (5.25-SSQD).7z”
6d68e316-97c0-11e9-ab10-fa163e9022f0 “Volkswriter 1.2 128k (1983) (5.25-160k).7z”
4152e167-19dc-11ea-9911-fa163e9022f0 “WordPerfect 4.1c (1985) (5.25-360k).7z”
2ddc9381-8683-11ed-aca0-0200008a0da4 “Zortech CPP 3.0r4 (1991) (5.25-1.2mb).7z”

All archives look fine, they can be downloaded, just checksums don't match. In some cases it's obvious that checksum is a nonsense (e.g. 4152e167-19dc-11ea-9911-fa163e9022f0 “WordPerfect 4.1c (1985) (5.25-360k).7z” have, literally, “WordPerfect 4.1c (1985) (5.25-360k).7z” in place of checksum), in many cases it's not clear if what's there is just part of checksum or maybe it's some exotic checksum…

But the majority of files in the archive match with checksums from the “download” links.

And an additional issue that is strange: for some reason some german versions of Internet Explorer are duplicated. Archives have the exact same names, but 3dc2a7c2-bbc3-8618-c39a-11c3a4e284a2, 3dc2ae15-c385-18c3-9a11-c3a4e284a2ef, 3dc2af5b-4818-c39a-11c3-a4e284a2c3a5, 3dc2b5c2-b7c2-a918-c39a-11c3a4e284a2, 3dc2bc11-c384-18c3-9a11-c3a4e284a2ef, 3dc2bd58-7918-c39a-11c3-a4e284a2c3a5, 3dc383c2-b537-18c3-9a11-c3a4e284a2ef, 3dcb9c76-c3ab-18c3-9a11-c3a4e284a2ef are in the “Beta Applications/PC” while 3d3e2c54-18c3-9a11-c3a4-e284a2c3a570, 3d40c2b6-c2a3-18c3-9a11-c3a4e284a2ef, 3d41c3bb-c3bd-18c3-9a11-c3a4e284a2ef, 3d43414f-18c3-9a11-c3a4-e284a2c3a570, 3d45c38f-1a18-c39a-11c3-a4e284a2c3a5, 3d47140f-18c3-9a11-c3a4-e284a2c3a570, 3d4859c2-b518-c39a-11c3-a4e284a2c3a5, 3d49c29d-c3b9-18c3-9a11-c3a4e284a2ef are in the “Beta Applications/PC (International)”. They have the exact same content, except for different disclaimer (winworldpc.com.txt file)..

Comments

  • Thanks for pointing out the errors. The checksums are manually entered in to the database, so I don't doubt there are errors. Yea, I'll find some small error in how I packed it, re-upload it, and might forget to update the checksum.

    Checksums were more important back when the mirrors were using plain HTTP. They are HTTPS now, so less issue of corruption in transit.

    I'll look in to those whenever I can find some time.
  • I used checksums to avoid re-downloading files that I already had… found out I have around 300GB of files already – more than half by size, yet not by number.

    And yes, it's obvious some files were changed over years… but some changes are… really strange for me.

    E.g. “Microsoft FORTRAN Optimizing Compiler 4.1 (5.25).7z” file downloaded long ago… today it's download 4863c2bf-c2a5-18c3-9a11-c3a4e284a2ef that gives you file “Microsoft FORTRAN Optomizing Compiler 4.1 (5.25).7z”… but why anyone would want to take correct name and change it to incorrect one?
  • According to what I have, it has been "Microsoft FORTRAN Optomizing Compiler 4.1 (5.25).7z" since 2010. 006cee668597227225b3ca00cc96adcb41bc2ced

    Note that the "download" name on the page and actual file name may differ.
  • Ah, right. Since I pulled it from collection of files I had before… it's possible that I have renamed it locally years ago, when I downloaded it, initially. But I definitely couldn't have changed content of file to match sha512 sum published on the web site… that's supposed to be cryptographically impossible.
Sign In or Register to comment.