- Boost - lists.boost.org

Re: [boost] Amazon SES
by Gavin Lambert 04 May '23

04 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

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

04 May '23

Le 2023-05-03 15:02, Sam Darwin via Boost a écrit : > Mailgun has been in production for many years. It costs $500 per month > to send smtp messages. It's a reasonable idea to try out another > service that's only 5% of the cost. Thanks for the detailed answer. From what is seen, it does indeed looks like amazon ses is sending emails with different message ids to each participant in the mailing list. Mailgun, on the other hand, was keeping the original message id in the email. > But at the same time, it's not impossible to switch back to mailgun, > if threading is partially disrupted, and we can't find a solution. Now that the source of the problem has been identified, i hope a solution can be found. Like Phil noted, mailgun is indeed very expensive. Best regards, Julien

1 0

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

04 May '23

On 5/4/23 10:20, Gavin Lambert via Boost wrote: > On 4/05/2023 18:34, Julien Blanc wrote: >> You're luckier than me. The current thread is semi-broken, but the >> "replacing the ML with a forum" one is completely broken for me. > > Sorry, that was sloppy reporting on my part. There was a large tree of > posts that were all together so I foolishly assumed that this was the > whole thread. > > There were *also* a bunch of disconnected replies earlier on, but since > these were completely disconnected they didn't "follow" the parent > higher in the list since I'm sorting by date. This meant they were out > of sight at the time I was thinking about this and writing my message. > >> I'm starting to wonder if we all receive the same emails (in the sense >> that they may not contain the same set of headers). For me, the >> initial message of Vinnie in the "replacing the ML with a forum" >> thread contains the following: >> >> Message-ID: >> <010f0187bde5599a-e76c27f2-225d-4d7e-b2c8-887e3a07ecb8-000000(a)us-east-2.amazonses.com> >> >> But Mashshall's answer, on the other side, contains the following >> headers : >> >> References: >> <010f0187bde530fb-ff3c1911-df7d-4399-b5d2-82f8062bf18c-000000(a)us-east-2.amazonses.com> >> In-Reply-To: >> <010f0187bde530fb-ff3c1911-df7d-4399-b5d2-82f8062bf18c-000000(a)us-east-2.amazonses.com> >> >> The "In-Reply-To" should match the Id of the initial message, for >> threading to work correctly. This is too obvious to be a client bug. >> Can someone check the Message-ID he has for the initial message of >> Vinnie in the "replacing the ML with a forum" thread? If it's >> different from mine, then that's probably the source of the issues. > > For me (via gmane), Vinnie's original message has: > > Message-ID: > <010f0187bde523cc-f60c3688-3843-44d4-b848-f37b2fd6a598-000000(a)us-east-2.amazonses.com> > > And Marshall's reply (did not thread properly) has: > > Message-ID: > <010f0187bde8bc2e-bebb94d1-2dfb-4165-ad8d-6af8e9b41381-000000(a)us-east-2.amazonses.com> > References: > <010f0187bde530fb-ff3c1911-df7d-4399-b5d2-82f8062bf18c-000000(a)us-east-2.amazonses.com> > In-Reply-To: > <010f0187bde530fb-ff3c1911-df7d-4399-b5d2-82f8062bf18c-000000(a)us-east-2.amazonses.com> > > Robert's reply (did thread properly) has: > > Message-ID: > <010f0187bf4b951d-1f6d7af8-3f2e-458b-9ffd-ca91ecfe04f9-000000(a)us-east-2.amazonses.com> > References: > <010f0187bde523cc-f60c3688-3843-44d4-b848-f37b2fd6a598-000000(a)us-east-2.amazonses.com> > In-Reply-To: > <010f0187bde523cc-f60c3688-3843-44d4-b848-f37b2fd6a598-000000(a)us-east-2.amazonses.com> > > These do indeed seem to be different from yours, which is unfortunate. > > It would be good if someone else not using gmane could check both > results as well (although that shouldn't be a factor, as I have other > groups that don't appear to be getting flattening). I'm reading the ML in Thunderbird. Vinnie's initial message has: Message-ID: <010f0187bde531b6-2cd00e40-57e1-476f-b30d-896ff8eab806-000000(a)us-east-2.amazonses.com> Marshall's reply: References: <010f0187bde530fb-ff3c1911-df7d-4399-b5d2-82f8062bf18c-000000(a)us-east-2.amazonses.com> In-Reply-To: <010f0187bde530fb-ff3c1911-df7d-4399-b5d2-82f8062bf18c-000000(a)us-east-2.amazonses.com> Message-ID: <010f0187bde8ccef-31042da0-3d20-4f4a-95e8-74cd3770cc89-000000(a)us-east-2.amazonses.com> Robert's reply: Message-ID: <010f0187bf4ba591-77f55894-1b8b-40d4-b8ac-0d6af2b9ba6d-000000(a)us-east-2.amazonses.com> References: <010f0187bde523cc-f60c3688-3843-44d4-b848-f37b2fd6a598-000000(a)us-east-2.amazonses.com> In-Reply-To: <010f0187bde523cc-f60c3688-3843-44d4-b848-f37b2fd6a598-000000(a)us-east-2.amazonses.com> For me, both Marshall's and Robert's replies, which are direct replies to the initial Vinnie's post, start their own threads with their subtrees of replies. The only descendant of the Vinnie's initial post is my own reply to it, which has this: Message-ID: <187be1c7608.280f.fb49792bc380c7e55e30f872b414f11c(a)gmail.com> In-Reply-To: <010f0187bde531b6-2cd00e40-57e1-476f-b30d-896ff8eab806-000000(a)us-east-2.amazonses.com> References: <010f0187bde531b6-2cd00e40-57e1-476f-b30d-896ff8eab806-000000(a)us-east-2.amazonses.com> So basically, every direct reply to the original post, except my own one, is a separate subtree (or a single out-of-tree message if there were no replies to it).

1 0