Dňa 22. októbra 2024 10:03:12 UTC používateľ Vsevolod Stakhov via mailop
<mailop@mailop.org> napísal:
>I don't still accept your arguments about encoding overhead. We **already**
>have a huge overhead of UTF8 for non ASCII texts. Do you probably suggest to
>get back to, let's say, KOI8 for cyrillic characters to save environment? It
>will give you like 100% improvement for sure.
I am not familiar with KOI-8, but i will guess, that it is something as
CP852 (or how it was named ~30 years ago). History shows, that
extendend ASCII (the Windows-1250 vs. ISO-8859-2 in our/my case)
was more harm than good. That was wrong way, but it was step in
good direction (allow to use national chars). UTF-8 is, perhaps not
perfect, but great solution to remove confusion from different
enhanced codes for the same char. Thus, UTF-8 is not overhead, but
another step in good direction.
Enhanced ASCII was good, when Internet was not as common,
thus computers from different nations/charsets did not meet (as
often). I remmember how hard was at start of this century use
MS Word documents in between Windows and MacOS...
When we can agree that (7b) ASCII is not enough for most of the
world, without 8BITMIME it nowadays looks:
write 8bit -> encode to 7bit -> transfer 8bit -> decode to 8bit -> show 8bit
With 8BITMIME, two steps can be skipped:
write 8bit -> transfer 8bit -> show 8bit
Do you see the overhead now?
>It is, but the world is very far from being perfect. About 15 years ago, I
>would have been on your side: "let's burn the old world and build a shiny new
>one!". It doesn't work in the real life. When I was working on some TCP
>related research, the problem of middleboxes was really a huge one (e.g. here
>is a nice article related to that:
>https://inl.info.ucl.ac.be/system/files/hotmb11-hesmans.pdf).
>
>We can't simply close our eyes on that in the email world either. I wish we
>could use only relaxed DKIM as simple was broken by design (of course, with
>shiny ed25519 and not dull RSA). I wish we could use only SMTPUTF8 as
>everything before was broken by design. I wish we could deprecate all
>encodings aside of UTF8. I wish we could bury all TLS versions but keep 1.3
>only. Finally, I wish I could even use IPv6 and forget about IPv4 and dual
>stack.
My intention is not to change the world. I am enough old to
know, that it is really hard job. Instead, i only want to name
things correctly -- deliverability is not broken by those using
8BITMIME, but by those who don't accept/support it (nowadays).
In other words, i consider forcing of encoding 8bit to 7bit
nowadays as workaround to comply with some mastodons,
not as solution of problem.
But yes, i agree that we cannot close eyes about that missing
support, but that doesn't mean, that we have silently to agree
with it.
>Who cares about those wishes after all? The Internet is full of legacy and we
>must accept it and try not to break E2E compatibility in the wild world of the
>middleboxes of all sorts.
There are multiple types of wishes, some are truly subjective
and another are based on technological progress. I consider
8BITMIME as the second and the 7bit should become only
historical thing.
regards
--
Slavko
https://www.slavino.sk/
_______________________________________________
mailop mailing list
mailop@mailop.org
https://list.mailop.org/listinfo/mailop
