Matija Nalis <[email protected]> writes: > On Mon, May 11, 2026 at 06:45:03PM -0400, Greg Troxel wrote: >> Well, it's not ok to send html mail at all. But I just got a non-spam, >> actual messages from a human, that other than being html, was not at all >> spammy. The HTML_FONT_SIZE_HUGE rule fired. > > OK... and why do you think it is a problem? To wit:
I happen to score that up a bit. That's my issue. It's a problem because it's asserting something about the message that is not true and makes no sense. That's a bug even if the default config only counts it for one millipoint. >> I can't post the whole message, but there is > > Well, perhaps you should create minimal representative example then. Thanks for pushing me to do that. I went through and trimmed and trimmed, checking each time, and converted identifiers to dummy values. The result is super short and it still fires. > At quick glance, Mail/SpamAssassin/HTML.pm (at least in 3.4.6 version > of SA) does not trigger on CSS style "font-size: *pt" at all, but > e.g. on (deprecated) HTML "<font size=*>" which is totally different > metric from "pt", and can range from 1 to 7 (with 3 being "default text > size"). Ah, that's what I semi-remember. Thanks for the doc pointer. I have attached my "hample" at end as application/octet-stream so it will arrive unmunged. Rerunning with -t, I get the following output: Content analysis details: (5.5 points, 1.0 required) pts rule name description ---- ---------------------- -------------------------------------------------- -0.0 RCVD_IN_DNSWL_NONE RBL: Sender listed at https://www.dnswl.org/, no trust [2607:f8b0:4864:20:0:0:0:1133 listed in] [list.dnswl.org] 1.9 SPF_FAIL SPF: sender does not match SPF record (fail) [SPF failed: Rejected by SPF record.] 0.0 SPF_HELO_NONE SPF: HELO does not publish an SPF Record 0.1 MSGID_PROD_OUTLOOK Message-Id from prod.outlook.com 1.0 HTML_FONT_SIZE_HUGE BODY: HTML font size is huge 0.1 HTML_MESSAGE BODY: HTML included in message 1.8 DMARC_REJECT DMARC reject policy 0.6 TXREP TXREP: Score normalizing based on sender's reputation The SPF and DMARC are because I've messed with identifiers, so we should ignore that. MSGID_PROD_OUTLOOK is a personal meta-rule building block. TXREP, well I've been messing with it. But the big deal is I'm still getting HTML_FONT_SIZE_HUGE. In the stderr with -D May 11 21:50:45.236 [29438] dbg: rules: ran eval rule HTML_FONT_SIZE_HUGE ======> got hit (1) SA 4.0.2, perl 5.42.2, NetBSD 10 amd64. sa-update nightly: rule files have mod time of 0500 EDT today. This setup otherwise works well.
ham.000
Description: Binary data
