Here are the steps I did to illustrate this related "strange" problem (bug?):
1. (I already had 7-zip v9.20 installed in the "C:\Program Files\7-Zip" folder.)
2. I downloaded 7-zip v9.20.04 using the link Igor provided:
(see " http://sourceforge.net/projects/sevenzip/forums/forum/45797/topic/4026469 ").
3. I created the "C:\Program Files\7-Zip\_v9.20 with bug" subfolder.
4. I moved v9.20 files "7z.dll" and "7zFM.exe" into that subfolder.
5. I extracted the (32-bit) v9.20.04 files "7z.dll" and "7zFM.exe" (NEW).
6. I moved those two NEW files into the ""C:\Program Files\7-Zip" folder.
Then, in keeping with the "Merry Christmas and Happy New Year" theme, I did this:
7. In a separate, empty folder I created "example.txt" (a new text file).
8. I put as the ONLY TEXT in "example.txt" the two letters "mc" (lowercase).
9. I saved the "example.txt" file.
10. I right-clicked on "example.txt" and chose 7-zip's "Add to archive…" option.
11. I named the new archive "last.7z" and set "Pavlov" as the password.
12. I right-clicked on "last.7z" and chose 7-zip's "Add to archive…" option.
13. I named the new archive "first.7z" and set "Igor" as the password.
14. I deleted the "example.txt" and "last.7z" files, leaving only "first.7z" there.
15. I opened the "first.7z" archive and successfully used "Igor" and "Pavlov" as passwords.
16. I checked the contents of the archived (ultra compression) "example.txt" file as OK.
17. I repeated steps 15 and 16 again, and everything was still OK.
18. I added to the "mc" text in "example.txt" the text "and hny" and saved it throughout.
19. I repeated step 15. The passwords were intact BUT the step 16 text was now CORRUPTED:
20. The text in "example.txt" ("mc and hny") had somehow corrupted (to 5 tiny "boxes")!
21. I am able to continually reproduce this issue. Do you get the same results?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2010-12-26
Here are some more details on this strange phenomenon that I just discovered:
If the same steps are taken (with respect to the filename and the sample text) …
… the "5 tiny box" corruption will occur even if "Store" (instead of "Ultra compression") is used;
… the "5 tiny box" corruption will occur even if NO passwords are used on the "first.7z" and "last.7z" archives;
… BUT … the "5 tiny box" corruption does NOT occur if the original text ("mc") is capitalized to "MC" instead.
This is really strange. Hopefully these descriptions can help isolate and repair this problem.
Can it be that ONLY this one test file will corrupt? - or can other files with a similar pattern also corrupt if changed in this way?
There's nothing like a good mystery…. *<:-)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2010-12-26
After further testing, this doesn't seem to be a 7-zip issue after all!
Rather, it seems to be something in the way Windows itself works (at least Windows XP and Windows 2000).
Just creating a text file with any name and trying to save therein the text "mc and hny" seems to later produce those 5 tiny boxes!
(Go figure - leave it up to me to coincidentally come up with this quirky-in-itself example around Christmas!) ;-D
Igor, I still think you have the best compression program out there! Thanks for the continued support!
I really enjoy using 7-zip and am grateful for your expertise in creating/maintaining it and your kindness in sharing it! :-)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I confirm this problem for Notepad 5.0.2140.1 (Windows 2000 SP4). Notepad 6.0.6001.18000 (Windows Vista SP2) is unaffected. I don't know whether it is a bug or rather an Easter egg, but Notepad seems to recognize "mc and hny" as Unicode-16 (so you see 5 unused Unicode symbols instead of 10 ASCII ones). Binary representation remains the same, so it is plainly Notepad's malfunction, not 7-Zip's one. There also are other strings that drive Notepad mad.
P.S. I have once developed a codepage recognition program myself. It is quite difficult, especially for languages with non-Latin alphabets. For example, Unicode endianness markers map onto Cyrillic letters which can (but rarely) stand together. There is no legal way to patch Notepad 5.x, so if you need to store such strings in TXT-files, pad them so that they consist of odd number of bytes.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Here are the steps I did to illustrate this related "strange" problem (bug?):
1. (I already had 7-zip v9.20 installed in the "C:\Program Files\7-Zip" folder.)
2. I downloaded 7-zip v9.20.04 using the link Igor provided:
(see " http://sourceforge.net/projects/sevenzip/forums/forum/45797/topic/4026469 ").
3. I created the "C:\Program Files\7-Zip\_v9.20 with bug" subfolder.
4. I moved v9.20 files "7z.dll" and "7zFM.exe" into that subfolder.
5. I extracted the (32-bit) v9.20.04 files "7z.dll" and "7zFM.exe" (NEW).
6. I moved those two NEW files into the ""C:\Program Files\7-Zip" folder.
Then, in keeping with the "Merry Christmas and Happy New Year" theme, I did this:
7. In a separate, empty folder I created "example.txt" (a new text file).
8. I put as the ONLY TEXT in "example.txt" the two letters "mc" (lowercase).
9. I saved the "example.txt" file.
10. I right-clicked on "example.txt" and chose 7-zip's "Add to archive…" option.
11. I named the new archive "last.7z" and set "Pavlov" as the password.
12. I right-clicked on "last.7z" and chose 7-zip's "Add to archive…" option.
13. I named the new archive "first.7z" and set "Igor" as the password.
14. I deleted the "example.txt" and "last.7z" files, leaving only "first.7z" there.
15. I opened the "first.7z" archive and successfully used "Igor" and "Pavlov" as passwords.
16. I checked the contents of the archived (ultra compression) "example.txt" file as OK.
17. I repeated steps 15 and 16 again, and everything was still OK.
18. I added to the "mc" text in "example.txt" the text "and hny" and saved it throughout.
19. I repeated step 15. The passwords were intact BUT the step 16 text was now CORRUPTED:
20. The text in "example.txt" ("mc and hny") had somehow corrupted (to 5 tiny "boxes")!
21. I am able to continually reproduce this issue. Do you get the same results?
Here are some more details on this strange phenomenon that I just discovered:
If the same steps are taken (with respect to the filename and the sample text) …
… the "5 tiny box" corruption will occur even if "Store" (instead of "Ultra compression") is used;
… the "5 tiny box" corruption will occur even if NO passwords are used on the "first.7z" and "last.7z" archives;
… BUT … the "5 tiny box" corruption does NOT occur if the original text ("mc") is capitalized to "MC" instead.
This is really strange. Hopefully these descriptions can help isolate and repair this problem.
Can it be that ONLY this one test file will corrupt? - or can other files with a similar pattern also corrupt if changed in this way?
There's nothing like a good mystery…. *<:-)
After further testing, this doesn't seem to be a 7-zip issue after all!
Rather, it seems to be something in the way Windows itself works (at least Windows XP and Windows 2000).
Just creating a text file with any name and trying to save therein the text "mc and hny" seems to later produce those 5 tiny boxes!
(Go figure - leave it up to me to coincidentally come up with this quirky-in-itself example around Christmas!) ;-D
Igor, I still think you have the best compression program out there! Thanks for the continued support!
I really enjoy using 7-zip and am grateful for your expertise in creating/maintaining it and your kindness in sharing it! :-)
I confirm this problem for Notepad 5.0.2140.1 (Windows 2000 SP4). Notepad 6.0.6001.18000 (Windows Vista SP2) is unaffected. I don't know whether it is a bug or rather an Easter egg, but Notepad seems to recognize "mc and hny" as Unicode-16 (so you see 5 unused Unicode symbols instead of 10 ASCII ones). Binary representation remains the same, so it is plainly Notepad's malfunction, not 7-Zip's one. There also are other strings that drive Notepad mad.
P.S. I have once developed a codepage recognition program myself. It is quite difficult, especially for languages with non-Latin alphabets. For example, Unicode endianness markers map onto Cyrillic letters which can (but rarely) stand together. There is no legal way to patch Notepad 5.x, so if you need to store such strings in TXT-files, pad them so that they consist of odd number of bytes.