May 2023 - Boost - lists.boost.org

Re: [boost] Amazon SES
by Sam Darwin 05 May '23

05 May '23

In retrospect it was an irresistible experiment to try out Amazon SES. The service seems to offer a large cost savings while also providing reliable mail delivery. However, here is a reply from the mailman-users list: "Amazon SES rewrites the Message-ID. Yep and for this reason I don't use it. The problem is worse in Mailman3 because it has a much more superior archive display which gets ruined when people reply to messages with different Message-Id headers and the threading can't be observed. I also had to use the anonymous mode with my list as Amazon didn't like me using random addresses in the From: header. I'm still looking for a tolerant mail gateway to use with Mailman." On Thu, May 4, 2023 at 5:27 PM Gavin Lambert via Boost <boost(a)lists.boost.org> wrote: > > Mere moments ago, quoth I: > > There's not too much problem if the first message id is rewritten > > (provided both mailman and SES agree on what it is, but that seems hard > > if mailman is earlier in the traffic chain, as appears to be the case). > > Actually, I take that back. Messages might be cross-posted to multiple > lists, in which case it's important that these all have the same message > id too. While *usually* they would all be Boost lists, that's not > necessarily the case, so the only correct choice is to not alter the > original message id. > > > _______________________________________________ > Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

1 0

Re: [boost] Amazon SES
by Gavin Lambert 05 May '23

05 May '23

Mere moments ago, quoth I: > There's not too much problem if the first message id is rewritten > (provided both mailman and SES agree on what it is, but that seems hard > if mailman is earlier in the traffic chain, as appears to be the case). Actually, I take that back. Messages might be cross-posted to multiple lists, in which case it's important that these all have the same message id too. While *usually* they would all be Boost lists, that's not necessarily the case, so the only correct choice is to not alter the original message id.

1 0

Re: [boost] Amazon SES
by Gavin Lambert 05 May '23

05 May '23

On 5/05/2023 04:40, Sam Darwin wrote: > Checking original messages locally in my inbox messages, it appears > mailgun doesn't rewrite the ID. SES rewrites the ID. Not an option to > disable it. I will open a discussion with mailman and see if they have > any suggestions. I think the only solution would be to get SES to stop doing that. Or to stop using SES if it can't. There's not too much problem if the first message id is rewritten (provided both mailman and SES agree on what it is, but that seems hard if mailman is earlier in the traffic chain, as appears to be the case). But for second and subsequent messages in the same thread, replies are often sent both to the list and to individual participants, and it's critically important that both of these copies have the identical message id both so that servers and clients recognise them as duplicates (and that one is safe to discard) and so that it doesn't matter which copy is replied to next. Changing the id is just poor netiquette, and it should only be done if the incoming email had a malformed id.

1 0

Boost::Regex support with ICU for unicode characters.
by Neha Patil 04 May '23

04 May '23

Hello, We are working with boost::regex library to have unicode support. I want to form a regex for validating unicode characters so I'm trying this with boost library. I am facing an issue with a boost library with ICU support so please suggest me how to proceed with the boost with ICU support. PFA for this. 1. *has_icu_builds* is used to enable icu support. [image: image.png] 2. on *b2 install* no icu is there, please let me know the procedure of how to enable icu. [image: image.png] Also I tried attaching the ICU library externally but it gives incorrect results for regex having character classes. PFA for this. [image: image.png][image: image.png] *NOTE *: Using 'Lo' character class as this class supports unicase alphabets. Kind Regards, Neha Patil.

1 1

Re: [boost] Q1 2023 Update: C++ Alliance's Efforts (PDF Attached)
by Daniela Engert 04 May '23

04 May '23

Am 04.05.2023 um 01:49 schrieb Gavin Lambert via Boost: > On 4/05/2023 03:55, Andrey Semashev wrote: >> The fact that the initial message is not part of the thread now is a >> relatively recent issue. This is the mailman archive from a year ago: >> >> https://lists.boost.org/Archives/boost/2022/04/index.php >> >> Threading is fine there. In fact, threading was working correctly as >> recently as in February 2023: >> >> https://lists.boost.org/Archives/boost/2023/02/index.php >> >> and has been broken in March: >> >> https://lists.boost.org/Archives/boost/2023/03/index.php >> >> If the system works (and worked) the way you describe, mailman was able >> to associate the initial message with the thread before and now it >> isn't. I suspect, this may have something to do with the headers that >> are added by SES to the messages it sends to the subscribers. > > Threading only sometimes works in the gmane view as well. Usually > it's not the mailing list itself at fault, but due to the mail client > used by the person posting -- or sometimes that they're replying to a > digest rather than the original mail, so the context headers are lost > (even if they edit it to look like it was a reply to a single post). > > I have noticed an apparent increase recently in the number of posts > that appear "flat" or semi-flat (still showing as a reply, but to a > grandparent post rather than the real parent post) -- but other recent > threads still do appear normally nested, so I don't think it's a > systemic issue. > I get the posts on the ML delivered as individual emails to my client (Thunderbird) and I can see the routing. Threading was fine until the end of February (through mailgun.net) and is broken since the beginning of March (through amazonses.com) The transition can be seen in the thread "Has testing on msvc been dropped?) in Tom Kent's posts. Dani -- PGP/GPG: 2CCB 3ECB 0954 5CD3 B0DB 6AA0 BA03 56A1 2C4638C5

1 0

Re: [boost] Boost.Compat
by Ruben Perez 04 May '23

04 May '23

> > > > So my suggestion is to create a separate library/repo, Boost.Compat > (boostorg/compat, namespace boost::compat) for them. > > Thoughts? (Be sure to read the discussion in the PR as well.) > I'm in favor of this. I agree with not allowing extensions, too. I think a light review should take place before inclusion. Regards, Ruben.

1 0

Re: [boost] Amazon SES
by Sam Darwin 04 May '23

04 May '23

Checking original messages locally in my inbox messages, it appears mailgun doesn't rewrite the ID. SES rewrites the ID. Not an option to disable it. I will open a discussion with mailman and see if they have any suggestions. On Thu, May 4, 2023 at 10:30 AM Andrey Semashev via Boost <boost(a)lists.boost.org> wrote: > > On 5/4/23 18:54, Sam Darwin via Boost wrote: > > The comments about Message-ID: , References: , and In-Reply-To: > > headers are very interesting. "The Amazon Simple Email Service (AWS > > SES) rewrites the Message-ID header of outgoing emails." > > Consider the very first email of a thread. It arrives at the mailman > > server with a Message-ID that mentions "gmail", for example, and is > > added to the Archives. The message is sent out to all list members, > > who see a new Message-id of > > ...010f0187dca41ec8(a)us-east-2.amazonses.com Later replies can set > > the In-Reply-To header to refer the "@us-east-2.amazonses.com" > > Message-ID. But the first archived message didn't yet have that ID, > > since it arrived directly at the server, not through SES. The mailman > > archives on lists.boost.org organize that first message in a separate > > thread, presumably based on the original Message-ID. > > Can we disable Message-ID rewriting? > > Do we know how mailgun works wrt. Message-ID and other headers? > > > _______________________________________________ > Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

1 0

Re: [boost] Amazon SES
by Andrey Semashev 04 May '23

04 May '23

On 5/4/23 18:54, Sam Darwin via Boost wrote: > The comments about Message-ID: , References: , and In-Reply-To: > headers are very interesting. "The Amazon Simple Email Service (AWS > SES) rewrites the Message-ID header of outgoing emails." > Consider the very first email of a thread. It arrives at the mailman > server with a Message-ID that mentions "gmail", for example, and is > added to the Archives. The message is sent out to all list members, > who see a new Message-id of > ...010f0187dca41ec8(a)us-east-2.amazonses.com Later replies can set > the In-Reply-To header to refer the "@us-east-2.amazonses.com" > Message-ID. But the first archived message didn't yet have that ID, > since it arrived directly at the server, not through SES. The mailman > archives on lists.boost.org organize that first message in a separate > thread, presumably based on the original Message-ID. Can we disable Message-ID rewriting? Do we know how mailgun works wrt. Message-ID and other headers?

1 0

Re: [boost] Amazon SES
by Sam Darwin 04 May '23

04 May '23

The comments about Message-ID: , References: , and In-Reply-To: headers are very interesting. "The Amazon Simple Email Service (AWS SES) rewrites the Message-ID header of outgoing emails." Consider the very first email of a thread. It arrives at the mailman server with a Message-ID that mentions "gmail", for example, and is added to the Archives. The message is sent out to all list members, who see a new Message-id of ...010f0187dca41ec8(a)us-east-2.amazonses.com Later replies can set the In-Reply-To header to refer the "@us-east-2.amazonses.com" Message-ID. But the first archived message didn't yet have that ID, since it arrived directly at the server, not through SES. The mailman archives on lists.boost.org organize that first message in a separate thread, presumably based on the original Message-ID.

1 0

Re: [boost] Q1 2023 Update: C++ Alliance's Efforts (PDF Attached)
by Vinnie Falco 04 May '23

04 May '23

On Wed, May 3, 2023 at 11:44 PM Julien Blanc via Boost <boost(a)lists.boost.org> wrote: > the "replacing the ML with a forum" That's not happening, no one is replacing the ML, and our efforts to renovate the Boost website have exactly one overarching principle: "Do not fuck with the mailing lists" The mailing lists must not be touched, they need to work exactly the way they did before (with working threading). Thanks

1 0