>
> No policy, but I'd be curious to know what the OP believes to be
> practically accomplished with signed files. Perhaps we're just talking
> about the official binaries? Or just checksums?


I'm only interested in the official binaries. The problem of determining
that the sources retrieved from the official master repository are the same
sources is something else entirely.

1. Integrity. I know the binary has not been modified in transit in some
form. Catalog signing, like the MD5 file talked about here, also
accomplishes this, provided that there is something that signs it, and so
on.

2. Identity. I know that the person claiming to be Bram Moolenaar (or Steve
Hall, or whomever) is certified to be that person by some certification
authority I already trust.

3. Authorship. Combining the previous benefits, I know the file is intact,
that Bram really is a person/org, and that he really produced this file.

4. Provenance. I know that the binary I got from vim.org actually
originated with someone who both controlled vim.org and also the private
cert for codesigning the binaries there. (This is only if vim.org supports
https, which it currently does not.)

5. UX benefits. I'm restricting this to Windows, since I have no idea of
the state of PKI/code signing/etc on Linux or MacOS. On Windows,
executables that are digitally signed are presented differently than
binaries which are unsigned.

6. Revokability. If the prior constraints do not hold true (due to a
systems failure, vulnerability, or loss of private key, for instance) the
certificate can be revoked immediately.

7. Individual revokability. If a particular binary suffers from a very bad
vulnerability, it can explicitly be pulled.

There's also a couple white-listing benefits, which are completely
ancillary.

8. Anti-malware benefits. Most AV engines (and in particular the one used
by MS, for instance in Security Essentials) are able to author whitelist
signatures for known good certs.

9. Reputation services (like Smart Screen for downloads in IE9). Over time,
these can provide actual trust benefits (like
http://www.hanselman.com/blog/UsingCodeSigningCertificatesToSignDownloadedMSIsAndBuildReputationWithIE9SmartScreen.aspx
 illustrates.)

In reality, my personal motivation is to get rid of that damn unsigned
dialog, but from an objective standpoint my motivations don't matter. :)

Philip

-- 
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php

Raspunde prin e-mail lui