Rspamd: improve SPF, DKIM and DMARC Symbol Weights #3913
Conversation
The order was chosen because it will reflect an ascending symbol weight with upcoming commits. I also added comments to explain when a symbol is applied.
I updated the symbol weights according to my new insights in #3690 to fix a bug pointed out by @2GetApp and to improve the logic itself. Previously, I reasoned about combinations of symbols that cannot exists, e.g., SPF allow, DKIM allow, DMARC reject. Removing these symbols and then reasoning about the rest is more appropriate. Moreover, I added `DMARC_POLICY_NA` and `DMARC_POLICY_SOFTFAIL` to the whole calculation. The issue description of #3690 I updated. I also added the Rust code I used to do and verify the calculations.
11fef33 to
9d828c5
Compare
polarathene
left a comment
There was a problem hiding this comment.
Can't comment on the actual config changes themselves, but here are some suggestion for possible improvements 👍
I noticed the changelog entries omitting the # prefix a few times, should probably correct that.
Co-authored-by: Brennan Kinney <[email protected]>
|
I have slightly adjusted the suggestion on the warning admonition you posted @polarathene. I hope you like it :) |
|
Oh my, what in the world happened to the links on the Rspamd page??? EDIT: I corrected this mishap in f3bfe32. Better merge fast, this is ludicrous. |
|
Documentation preview for this PR is ready! 🎉 Built with commit: d327f5c |
That was due to my review suggestion being applied from a prior PR, but getting approved without the corresponding change through the rest of the document 😓 My bad, I should really try find time to revive my old work on integrating |
|
I actually found the issue because of the line I use in VS Code (that you also linked to @polarathene), but I'm afraid this would cause a bunch of work when implemented as your MD may not conform very well with the standards.. |
PR - docker-mailserver/docker-mailserver#3833 Also mentioned in docker-mailserver/docker-mailserver#3913 Signed-off-by: Aldo Maria Vizcaino <[email protected]>
Description
I updated the symbol weights according to my new insights in #3690 to fix a bug pointed out by @2GetApp and to improve the logic itself.
Previously, I reasoned about combinations of symbols that cannot exist, e.g., SPF allow, DKIM allow, DMARC reject. Removing these symbols and then reasoning about the rest is more appropriate.
Moreover, I added
DMARC_POLICY_NAandDMARC_POLICY_SOFTFAILto the whole calculation.The issue description of #3690 I updated. I also added the Rust code I
used to do and verify the calculations.
Please review commit by commit.
Successor of: #3726
Fixes #3690
Type of change
Checklist
docs/)CHANGELOG.md