Topic: First signature of this email is not valid

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi all,
      I get this first signature of this email is not valid. I am getting this in linux. This is when I'm using my PGP signature to sign. The way I do it is I choose a chunk of text, right-click, choose FirePG > Sign & then sign it. I am using GNU Privacy Assistant (GPA) and have the following things done there :-

The key has no expiry date, it has Ultimate Ownership Trust and it is fully Valid. Then why does FirePG says its invalid?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: http://firegpg.tuxfamily.org

iD8DBQFGSNhylQ1T+3KaixcRArKFAJ9HWIkuoG6dypWStv57q+/rOoDwcgCeKMrO
qvhTQSPzY/CzCwrNkaHQTjw=
=yrL/
-----END PGP SIGNATURE-----

Re: First signature of this email is not valid

It's strange you're banned... Maybe our anti-spam... Fixed.

Yes, some time, the sign failed. WE got a big mail with a lot of explanations... We will see.

Re: First signature of this email is not valid

I ran a few tests

1. Didn't work. Checked this forum.
2. Signed whole message. Didn't work.
3. Does one of the algorithms truncate ending spaces or other "hidden" characters? Signed a part in the middle of the message. Didn't work.
4. Is it the word wrap? Gmail allows word wrap in plaintext email during composition and then inserts its own carriage returns (I think after the last word such that each line is 80 characters or less). I tested just my contact info, every line has a carriage return, and then the signature verified.
5. In rich text mode, which doesn't force wordwraps, the above tests all work.
6. In basic html mode, which is inherently plaintext, forcing wordwraps, tests 1 - 4 didn't work.
7. In rich text mode, signing and then encrypting also works.

This thread has some additional reports. The first post advises composing in plaintext, which the above tests indicate is the wrong advice. Use rich text. You don't have to use the bad features of rich text, just use it because it preserve the data string.

http://firegpg.tuxfamily.org/forum/viewtopic.php?id=3

Hope this helps.

Last edited by Niels (2007-05-21 16:33:03)

Re: First signature of this email is not valid

I went over and read what they said about Rich Text.  Please enlighten me because the only options I see for the WebMail services are the following:

Gmail:
Choice of default (assume plain text) and UTF-8

Hotmail:
I can't find any choices so the default is probably plain text.

Yahoo:
plain text or color & graphics.  I assume putting it in the color & graphics mode allows you to embed graphics and thus it becomes HTML format.  Since plain text is safer, that is what I use.

AOL / Netscape:
I can't see any choice, but since I didn't want to set it to not view or compose all email messages in separate windows, I abandoned it.  At the time I wanted to use the Tools -> FireGPG for every message check because I was uncertain whether the GMail buttons were causing the problems. I will check it again in a few days now that I have it set not to open Window for every message read or composed to see if that makes a difference.  I am intentionally using the web mail services in the way most people use them - with the defaults.  Those are the defaults, and I didn't want to contend with a right click screwing up the results.

I do need to say that I have done EXHAUSTIVE tests with the first three Webmail services and one POP mail account using Thunderbird (it is the only MUA I use that has support for INLINE).  The signing results have been dismally poor at best.  You can all download the results of the test by pointing the browser to:

http://www.securemecca.com/FireGPG.zip

This file is HIDDEN.  In other words almost everything at those web pages are hidden, so don't go the host itself and expect to maneuver around and find the file.  I haven't got around to building the web site yet.

I was ready to say that it must be a browser thing, but if it is, then why are ALL of my tests of encryption okay?  I must admit that I am signing with a SHA-512 digest, and the encryption may be using only SHA1, but lots of other people are having problems signing using SHA1.  Even worse, there are times that the verify fails or succeeds with FireGPG, and it does the converse (succeeds or fails, respectively) when I copy the mail message into a file and verify it from the command line (gpg --verify).  That does NOT instill confidence.  I am working from Fedora Linux, so the RTF (I assume you mean Rich Text Format - now if you mean UTF-8 that may mean something else) is almost meaningless.  You can't count on sending your messages to your POP or IMAP mail friends and have them do anything with them. There should be no difference between CR+LF versus just LF either since in making a check sum, gpg skips both of those characters.  Look at my tests in detail and maybe you can deduce something from it.  I am still searching for a pattern, and you have ALL of the short sign test email messages (I edited off the headers saved out by Thunderbird) to look at to make up your own mind.  I took the liberty of converting all LF -> CR+LF for the Windows users.  EMail sends with CR+LF anyway, not with LF.

I am NOT going to change my digest algorithm choices.  Further, All I know is that signing is OUT for FireGPG for now.  I can't even verify over 75% of my messages.  Even worse is what happens when I select the text and paste it into a file.  That becomes baffling when what verified in FireGPG doesn't verify from the command line, or what didn't verify in FireGPG does verify on the Command Line.

Asking for RTF is out of the question.  There is no native RTF (do you mean UTF-8?) support on Linux.  I don't want fancy messages verified.  Plain text is fine.  More to the point, since ALL of my encryption tests passed with flying colors, it makes the incosistent results of the sign tests even more baffiling.  I will do a few more tests with my AOL/Netscape webmail account (private UID addition to my keys) in the next few days and make it available at the same web site as AOL_FireGPG_SignTest.zip.

I really the FireGPG authors do need to make it plainly apparent that you should use INLINE signing at least, and preferably INLINE encryption (although web mail does handle OpenPGP/MIME encryption okay - the OpenPGP/MIME signing ends up with results that are all over the wall.  By that I mean one email service makes attached files, others embed things, etcetera. That makes OpenPGP/MIME impossible to use.  I figured it would.  Something needs to be said to users that they are primarily using INLINE (which I have no problem with even if I am using Evolution or Mac's mail app since I can save them out to a file and verify manually if I need to).

Be back in a few days ...

Re: First signature of this email is not valid

First, I must say right at the start that the Signing problems are NOT the fault of the FireGPG developers and I don't think they can do anything about it either. This will probably be my last post on this subject.  Here are the two files with fairly exhaustive signing tests:

http://www.securemecca.com/FireGPG.zip
http://www.securemecca.com/AOL_FireGPG_SignTest.zip

Paste these URLs DIRECTLY into your browser.  I don't have time to make a web page to point to data that will eventually be dated and removed.  We have too many stale links and outdated pages on the Internet as it is without adding some more. One surprising result is that EVERY test I had in sending signed messages from my hhhobbit7_GNAT_netscape.net email account to my hhhobbit_BAT_securemecca.net (POP mail) WORKS!  It doesn't just work in Thunderbird (where it always complains about the AOL tack-on). I saved the messages from Evolution which does NOT understand INLINE and just sees the message as text.  Every one of those messages also verified manually in a file. My version of Evolution only does OpenPGP/MIME; I can't speak for newer versions of Evolution.  So when I see any of the files with the *2tbd* I am actually saving them  from Evolution which makes no attempt to interpret anything.  I do strip off the headers, but so does gpg.  I did do some tests with and without the stripping and it makes no difference to the results (which it shouldn't).  It is the fact that the web interface itself is somehow changing what is handed to be signed and what is actually sent, and what is copied from the browser into the vim editor that I use to paste the email messages using X.

Signing has TONS of problems!  What I suspect is happening is that the WebMail program or FIrefox is giving FireGPG one thing which it faithfully hands off to gpg which signs it.  Additional CR+LFs don't pose a problem because GPG ignores them in making the check sum.  So what is happening is the WebMail (Firefox) is handing FireGPG one thing (which may include hidden characters), and what it sends is something else.  I am using Fedora Core 3 Linux with everything updated except for Evolution. I suspect it is some sort of hidden characters that you don't see.  The fact that it is always working from AOL to my POP mail account all the time seems to rule out Firefox itself, but that may not be correct.  All I know is that everything I have tested in signing from AOL / Netscape to my POP email account worked.

Okay, then why does signing not work when encryption works with no problems?  The reason why is that no matter what the Firefox browser / WebMail hands off to FireGPG, that is in turn encrypted and FireGPG will INCLUDE those funky characters if any in the encryption. What ever was originally there is COMPLETELY replaced by the encrypted test.  I don't know whether it is Firefox, the WebMail client, or what that is doing the swap of what is signed from what is sent, or even when it is happening, but it is taking place!  Given the random results of the verifying, it makes me suspect even more the WebMail itself over Firefox.  If it was Firefox, it would have more consistency.

Oh yes, the last test wasn't complete.  HotMail blocked all of my OpenPGP signed email from AOL / Netscape.  I am unsure whether it is the OpenPGP MIME marking, that AOL is now being blocked, or what. HotMail and MSN people need to realize that frequently email that is sent to their accounts is being blocked.  Sometimes when that happens, I can send the very same email message again with no problems..

For now my recommendation is to use FireGPG for INLINE asymmetric encryption only.  Further, I don't think the signing problems are due to anything the FireGPG developers have done.  In other words, I believe it is out of their control.  I may be wrong on that, but I have worked with it enough that is my final conclusion.  On the other hand, if they have released the code itself, I will take a look at it.  Without that anything else I would say is nothing more than a guess.

Finisez!

Re: First signature of this email is not valid

I do some test with bigs lines : that not a problem. I suspect now the \n and \r\n... I will do some test, we will make a INLINE who works !

Re: First signature of this email is not valid

Correct me if I'm wrong, but it seems the moral of the story is that FireGPG can't be used from within GMail for signing, end of story?  I've tried using Gmail to send, receive, and forward within Gmail and to and from other web and non-web clients, and my conclusion is that every message that touches Gmail becomes non-verifiable.  (This is not the case with encrypted messages--for some reason they can be decrypted and verified.)

If this can't be fixed, my suggestion to the developers of FireGPG is to remove the deceptive "Sign" button from the Gmail interface.  It's simply pointless to have a button whose purpose is thwarted in 100% of cases.

Does anyone have a different experience?  I'm using Firefox 3 on Ubuntu, and I wouldn't be surprised if mileage varied for people using different platforms.

Joe

Re: First signature of this email is not valid

No - in most of case (simples texts) it's works, from firegpg to firegpg. Next we have to be compatible wink

Re: First signature of this email is not valid

For some reason this is not what I have experienced--all the emails I have sent and received through GMail and signed and verified through firegpg have been recognized as invalid.  It doesn't matter if I use rich text, plain text, new GMail, old GMail, send to GMail or from GMail, if I sign it outside GMail and then paste it into GMail, etc.---it fails in 100% of cases.  It does work with Yahoo Mail and other web mail browsers though.

Is there some trick I should be trying to make it work?  For the moment I'm resorting to not using GMail if I want to sign something.

Re: First signature of this email is not valid

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Sorry to keep posting, but I continue to test this out and see what works and what doesn't.  I've concluded that, on my platform at least (Ubuntu Linux, Firefox 3), verifying a non-encrypted message using FireGPG almost never works, whether in GMail, another webmail client, or even a plain text file being viewed through Firefox 3.  Verification fails regardless of plain-text or wysiwig editing and even if the webmail client does not strip whitespace.  I haven't been able to figure out why.

Even the simplest possible test case, signing a .txt file, saving it in a text editor, opening it in Firefox, and verifying using FireGPG, does not verify, whereas the same text file verifies correctly from the command line or from within the text editor (gedit).  Whitespace is displaying correctly in Firefox.  Strangely, if I copy the same text, whether from my plain .txt file or from Gmail, from Firefox into gedit, gedit correctly verifies the exact same text that FireGPG has failed to verify, and I can save the same text and verify it from the command line using GPG.

Inexplicably, this message that I'm posting now does verify correctly.  I'm not sure why this text box is treated differently from a plain text file or an HTML file.  Could it be a problem with how Firefox 3 handles white space when rendering web pages?  Do others have this same problem in Firefox 3?

Joe Hill (0D3CE986)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: http://getfiregpg.org

iD8DBQFIWDBhZOor8Q086YYRAkysAJ9EALfJUbuoHbBzOMkOWN+i6wFcBQCdE7u5
3yNp3UHEf3nk9Tepl7Z7X9Q=
=NTYA
-----END PGP SIGNATURE-----

Re: First signature of this email is not valid

Hum. Yes it's very strange. I will investigate...

Edit: I just check a mail I just send. It's was valid. But I will have a look for your problem anyways wink

Last edited by the_glu (2008-06-17 23:01:07)

Re: First signature of this email is not valid

To follow up, the message I just posted above correctly verified when it was in the text box before being submitted, but now that it's being displayed as HTML, it no longer verifies.  The white space (double spaces between sentences, etc.) is displaying correctly.  If I copy and paste it into gedit, it verifies correctly.  And if I copy and paste it into FireGPG's "Text Editor" window, it also verifies.

Bottom line: as far as I can tell, when using my version of Firefox 3/Ubuntu, verifying signed, unencrypted text from a web page using FireGPG simply doesn't work.  (The problem isn't the hyperlinks, since I have the same problem when I load a plain text file.  And verifying/decrypting encrypted messages does work.)

Re: First signature of this email is not valid

Yop. It's probably the link, our cleaning function have probably a problem wink