#haraka

/

      • _smf_
        Hahaha. I'm fucked then ;-) I'm still using iojs in quite a few places.
      • ultimatt
        since they don't support lexical scope (let/const) w/o 'use strict' being enabled in the file
      • EyePulp_
        Have y'all ever discussed switching to yarn from npm?
      • ultimatt
        I think if 'use strict' had been enable everywhere (I do, when I notice/remember) it may have just passed
      • EyePulp_
        I've run into issues w/npm under node 8.x/npm 5.x and switching to yarn has been a nice improvement in numerous ways (including docker / CI/CD setup speeds)
      • ultimatt
        we haven't here, but I've played with yarn a bit due to the "npm 5.3 it a turd monster" issues. Sadly, yarn has it's own share of edge cases.
      • EyePulp_
        Ah, hadn't heard about yarn problems.
      • ultimatt
        Hah, nobody does until you still using it. Then you cut yourself on the edge cases and find 'em in the yarn repo
      • syshero joined the channel
      • I found it to be trading one set of problems for another, so stopped playing with it. for now.
      • EyePulp_
        just finished reading through #2073 ultimatt -- LGTM. =) Just remind me to bring a fresh set of pants when that PR merges. It's a whopper.
      • _smf_
        LOL
      • I'd say it LGTM too.
      • ultimatt
        Yeah, it's YUGE. But I don't expect much fallout. The stuff it does change (scope of variables) is really well covered by lint tests, and at least 3 sets of human eyes thus far
      • _smf_
        TBH - there's no way that human eyes are going to spot anything as it's so large and due to the nature of the change, lacks context to be able to where the variables are called etc.
      • ultimatt
        for sure, but, and I appreciate this about eslint, in those cases, it just did nothing
      • and left them for me to fix manually
      • _smf_
        Ok
      • ultimatt
        and surely that's why there haven't been tons of "OMG, your --autofix broke my stuff" issues with eslint
      • b/c yes, I did check. ;)
      • _smf_
        :-)
      • EyePulp_
        yeah, the --fix is impressively hands off at times. Nice tool/
      • _smf_
        I say we should just merge it and then test the crap out of it.
      • I'll even upgrade the spamtrap to HEAD - that should give it a good exercise....
      • ultimatt
        +1
      • and start a quiet period...
      • EyePulp_
        "Hey infant child, go run this marathon"
      • _smf_
        *nod*
      • ultimatt
        (after I finish up one more PR. ;-))
      • EyePulp_
        Your spamtrap sounds brutal.
      • _smf_
        It kinda is. Logging is *impossible* on it. I basically have it set at LOGEMERG
      • And I had to tune the OS and Haraka to handle that amount of traffic.
      • Some of the stuff I added to the Wiki
      • ultimatt
        Funny you mention that, I'm remembering back to the turn of the century when I created a "special mail server" that we could use DNS to route to when a customer was getting DDoS attacked. It just routed everything to /dev/null, and qmail had no problems keeping up with 50,000 concurrent deliveries on a dual P3 600, so long as you turned off tthe logging
      • _smf_
        Oh - I also had to disable /var/log/maillog entirely.
      • ultimatt
        with logging enabled, way, way way less
      • disks were so so so slow....
      • _smf_
        I can enable logging, but even with LOGNOTICE - the log files get massive really quickly.
      • And it's only got 24Gb SSD in it.
      • ultimatt
        we didn't SSDs back then. ;)
      • _smf_
        *nod*
      • Hahahah - I just checked the rspamd stats
      • Messages scanned: 40398134
      • I've no idea when I last reset them....
      • Messages with action reject: 38876968, 96.23%
      • bmonty joined the channel
      • godsflaw has quit
      • syshero has quit
      • chaosmaker joined the channel
      • chaosmaker joined the channel