Discussion:
UEFI boot basics
(too old to reply)
Paul Edwards
2023-11-28 00:42:10 UTC
Permalink
Hi James.
Any idea why your reply appears as a new thread?
Oh, that actually happened, did it? In google
groups my posts are appearing in the same thread
normally.

That initially surprised me, but then I thought
it was logical for situations where the reply is
not to anyone specific.

Bottom line is that I am using this crude program:

https://sourceforge.net/p/pdos/gitcode/ci/master/tree/src/pdpnntp.c

(which I wrote) since google groups became read-only.

I don't yet know how to construct a reply properly,
but even if I did, google groups isn't showing me
the raw headers (it used to), so I can't extract
the information I need.

I could get the raw headers via eternal september,
but that is more laborious. And I still don't know
the rules, regardless. I assume it's a simple matter
of getting the message-id from the original message
and then constructing some sort of in-reply-to
header line? If someone can confirm that, I'll try
to do a proper reply for you to confirm.
mingw64 has what is required.
As does UCX64.
Well, my development environment is Unix so I am not sure that either
would fit. More importantly, I think that I can already control what
goes in to the finished executable and have tried stripping out the
library code, it's just that there's an unresolved problem with the UEFI
firmware not liking the result.
Can you not run mingw under Wine to confirm that
you get a better result when you use standard
Windows calling convention and standard PE
executables?

I don't know if mingw64 can be built as a standard
Unix appication, but the tools in UCX64 are all
standard C90 applications except for cc64 which uses
a little bit of C99 (long long), so you need a compiler
with a little bit of C99 support (which you certainly
have), and then you can build native Linux tools that
produce proper PE executables instead of ELF-kludged.

And Wolfgang is probably using Windows so would probably
be better off going a route that doesn't involve the
Linux kludge.
Grant Taylor
2023-11-28 01:38:55 UTC
Permalink
Post by Paul Edwards
Oh, that actually happened, did it? In google
groups my posts are appearing in the same thread
normally.
Yep, your message is appearing as a new thread.
Post by Paul Edwards
That initially surprised me, but then I thought
it was logical for situations where the reply is
not to anyone specific.
Messages are to the newsgroup(s). -- It's possible to send a single
message to multiple newsgroups in Usenet. I don't know if Google groups
allows that or not.
Post by Paul Edwards
(which I wrote) since google groups became read-only.
Oh ... Google Groups is far from read-only in general. I think there
are some newsgroups that are now read-only in Google Groups UI. But the
thousands of messages from Google Groups flooding other newsgroups (as
of a few days ago) are evidence of the read-write nature of Google
Groups Usenet gateway.
Post by Paul Edwards
I don't yet know how to construct a reply properly,
but even if I did, google groups isn't showing me
the raw headers (it used to), so I can't extract
the information I need.
This is the 2nd or 3rd reference to Google Groups not making headers
available. That action really disappoints me. Yet more devolution of
Google Groups.
Post by Paul Edwards
I could get the raw headers via eternal september,
but that is more laborious. And I still don't know
the rules, regardless.
What rules are you thinking of / concerned about?
Post by Paul Edwards
I assume it's a simple matter
of getting the message-id from the original message
and then constructing some sort of in-reply-to
header line? If someone can confirm that, I'll try
to do a proper reply for you to confirm.
Yes.

References: <uk3d11$3vnan$***@dont-email.me>

Where "<uk3d11$3vnan$***@dont-email.me>" is the Message-ID: of the message
that I'm replying to.

References: is a Usenet construct while In-Reply-To: is more of an email
construct. That being said, it is quite common for messages to have both.

Both References: and In-Reply-To: include the Message-ID: of the message
that the reply is directly related to. One, I don't remember which, can
include more history like reply to a reply to a reply to a reply to an
original post type thing. -- Sorry, I can't be arsed to look up the
proper RFC for specifics.
--
Grant. . . .
Paul Edwards
2023-11-28 05:46:20 UTC
Permalink
Hi Grant.
Post by Grant Taylor
Post by Paul Edwards
Oh, that actually happened, did it? In google
groups my posts are appearing in the same thread
normally.
Yep, your message is appearing as a new thread.
Ok, is this one properly threaded?
Post by Grant Taylor
Post by Paul Edwards
(which I wrote) since google groups became read-only.
Oh ... Google Groups is far from read-only in general. I think there
are some newsgroups that are now read-only in Google Groups UI. But the
thousands of messages from Google Groups flooding other newsgroups (as
of a few days ago) are evidence of the read-write nature of Google
Groups Usenet gateway.
Ok, well, I can't click "compose" on comp.lang.c,
and in alt.os.development I can click "compose"
but last time I tried that I was asked to do a
captcha thing and then nothing happened. So I don't
want to waste my time fighting that anymore. I have
a news reader to write! And especially to get it to
fit into the Fidonet style, which is why there is an
"X-Comment-To".
Post by Grant Taylor
Post by Paul Edwards
I could get the raw headers via eternal september,
but that is more laborious. And I still don't know
the rules, regardless.
What rules are you thinking of / concerned about?
The one you clarified - I am required to provide a
"References", but not a "In-Reply-To".
Post by Grant Taylor
Both References: and In-Reply-To: include the Message-ID: of the message
that the reply is directly related to. One, I don't remember which, can
include more history like reply to a reply to a reply to a reply to an
original post type thing. -- Sorry, I can't be arsed to look up the
proper RFC for specifics.
No problem, since I have no intention of implementing
it anyway. :-)

However, I do intend to, and I believe I have, done
the "References", so now that I know there is an issue,
and now that I know the rules, I will hopefully do the
right thing going forward.

BFN. Paul.
Grant Taylor
2023-11-28 14:56:26 UTC
Permalink
Post by Paul Edwards
Hi Grant.
Hi Paul,
Post by Paul Edwards
Ok, is this one properly threaded?
It appears to be.
Post by Paul Edwards
Ok, well, I can't click "compose" on comp.lang.c,
and in alt.os.development I can click "compose"
but last time I tried that I was asked to do a
captcha thing and then nothing happened.
Sounds like specific groups may be read-only while other groups are
read-write.
Post by Paul Edwards
So I don't want to waste my time fighting that anymore.
Fair enough.
Post by Paul Edwards
I have
a news reader to write! And especially to get it to
fit into the Fidonet style, which is why there is an
"X-Comment-To".
I'm not familiar with the X-Comment-To: header.

Though being a header starting with X- it's eXperimental. That being
said, a number of things abuse eXperimental as production.
Post by Paul Edwards
The one you clarified - I am required to provide a
"References", but not a "In-Reply-To".
ACK
Post by Paul Edwards
No problem, since I have no intention of implementing
it anyway. :-)
Fair.

N.B. save for supporting HTML, which is not required anyway, news and
email are quite similar. So if you write a client for one, you have
much of the client code for the other too. If you ever care.
Post by Paul Edwards
However, I do intend to, and I believe I have, done
the "References", so now that I know there is an issue,
and now that I know the rules, I will hopefully do the
right thing going forward.
ACK

Learning at it's simplest form. You were unaware of something, you
became aware of it, and you became aware of how to overcome the issue.
<ASCII thumbs up>.
--
Grant. . . .
wolfgang kern
2023-11-28 04:40:40 UTC
Permalink
On 28/11/2023 01:42, Paul Edwards wrote:
...
Post by Paul Edwards
mingw64 has what is required.
...
Post by Paul Edwards
And Wolfgang is probably using Windows so would probably
be better off going a route that doesn't involve the
Linux kludge.
Yes, I use windoze for posting and games.
But for programming I use my own disassembling hex-edit
where only bare metal facts can happen at all.
I learned how to manually create short PE-files with that,
so all I now need to know is what all must be in efi.bin.
__
wolfgang
Scott Lurndal
2023-11-28 15:04:46 UTC
Permalink
Post by Paul Edwards
Hi James.
Any idea why your reply appears as a new thread?
Oh, that actually happened, did it? In google
groups my posts are appearing in the same thread
normally.
That initially surprised me, but then I thought
it was logical for situations where the reply is
not to anyone specific.
I don't yet know how to construct a reply properly,
but even if I did, google groups isn't showing me
the raw headers (it used to), so I can't extract
the information I need.
You have an NNTP application, update it to use
grab the headers.

You need to provide a Message-ID: header, and
update the References: header to append the message
ID of the message you're replying to.

See the RFC for NNTP for details.
Paul Edwards
2023-11-28 21:08:44 UTC
Permalink
Hi Scott.
Post by Scott Lurndal
You have an NNTP application
You can euphemistically call it that.
Post by Scott Lurndal
update it to use grab the headers.
I have that already. I just need to keep changing
the number until I find the right article.

The long-term plan is to ingest this into Msged,
a Fidonet application.

BTW, I was going to use UUCP on Eternal September
for that purpose. I don't yet know the protocol,
but I do get a userid/password prompt, and when I
enter my userid and password, Eternal September
terminates the connection. I tried some other userid
and passwords from examples in case this was not the
place you are supposed to enter your real userid
and password, but they didn't work either.

I do have a vague recollection that you need to
specifically request UUCP access, but I searched
and searched and couldn't find mention of such a
requirement anywhere.

Perhaps after entering the password I'm meant to
send something else very quickly or something,
and that's why it's terminating.
Post by Scott Lurndal
You need to provide a Message-ID: header, and
Eternal September, running INN I believe, automatically
provides that for me. From the Fidonet-like side, I have
no interest in controlling the Message-ID, so that's great.
Post by Scott Lurndal
update the References: header to append the message
ID of the message you're replying to.
Ok, so it's supposed to be append, not replace. That's
probably what Grant was alluding to before.

But I probably can't support unlimited length
headers in Msged, so may need to strip that
down.

Another thing is that Fidonet allows paragraphs -
basically long lines. That culture will need to
change for interaction with Usenet too.
Post by Scott Lurndal
See the RFC for NNTP for details.
Interestingly that (NNTP, RFC 977):

https://www.w3.org/Protocols/rfc977/rfc977.html

didn't have it.

But it was in USENET, RFC 850:

https://www.w3.org/Protocols/rfc850/rfc850.html

(aside to Grant - thanks for that info on the X-
being experimental - didn't know that)

BTW, it is fantastic to be back (since 1997) using
micro-emacs to write messages. But I really want
the Msged interface I was using back too. Now that
I have an operating system to run it under. Before
I was using OS/2.

BFN. Paul.
Scott Lurndal
2023-11-28 21:29:15 UTC
Permalink
Post by Paul Edwards
Hi Scott.
Post by Scott Lurndal
You have an NNTP application
You can euphemistically call it that.
Post by Scott Lurndal
update it to use grab the headers.
I have that already. I just need to keep changing
the number until I find the right article.
I assume you're familiar with RFC 3977.

Use the OVER command to get the the message-id
header and References header for the current article.

Then use XHDR (or the newer HDR/LIST HEADERS) commands.
Post by Paul Edwards
Ok, so it's supposed to be append, not replace. That's
probably what Grant was alluding to before.
Prepend, not append, I believe., e.g. here are the
headers for your post:

Path: not-for-mail
From: Paul Edwards <***@gmail.com>
Newsgroups: alt.os.development
Subject: Re: UEFI boot basics
Date: Tue, 28 Nov 2023 21:08:44 -0000 (UTC)
Organization: A noiseless patient Spider
Lines: 74
Message-ID: <uk5ksr$e7da$***@dont-email.me>
References: <uk3d11$3vnan$***@dont-email.me> <ikn9N.30033$rx%***@fx47.iad>
Injection-Date: Tue, 28 Nov 2023 21:08:44 -0000 (UTC)
Injection-Info: dont-email.me; posting-host="35c6e3a925efc030fe95aa1921b6c7ba";
logging-data="466346"; mail-complaints-to="***@eternal-september.org"; posting-account="U2FsdGVkX1/1qQ9OGF3agDIe8Ha3B+ckhY8dpEnMF30="
Cancel-Lock: sha1:Pj1/OA2hggwl6/+qaOjw3zW/O8E=
X-Received-Bytes: 3143
Post by Paul Edwards
But I probably can't support unlimited length
headers in Msged, so may need to strip that
down.
Another thing is that Fidonet allows paragraphs -
basically long lines. That culture will need to
change for interaction with Usenet too.
Post by Scott Lurndal
See the RFC for NNTP for details.
https://www.w3.org/Protocols/rfc977/rfc977.html
didn't have it.
try 3977.
Paul Edwards
2023-11-29 00:52:17 UTC
Permalink
Hi Scott.
Post by Scott Lurndal
Post by Paul Edwards
I have that already. I just need to keep changing
the number until I find the right article.
I assume you're familiar with RFC 3977.
Incorrect assumption.
Post by Scott Lurndal
Use the OVER command to get the the message-id
header and References header for the current article.
Then use XHDR (or the newer HDR/LIST HEADERS) commands.
Any such thing needs to be written. So far it
hasn't been written. The software in question is here:

https://sourceforge.net/p/pdos/gitcode/ci/master/tree/src/pdpnntp.c

My priority is to get that working on z/PDOS,
not add more bells and whistles to the PDOS/386 version.

So until then, the software requires me to choose a
single number. And I update a .bat file to specify
that number. And then if I get the number wrong, I
poweroff and do a fresh connection to Eternal September
(using qemu to point COM1: to there).
Post by Scott Lurndal
Post by Paul Edwards
Ok, so it's supposed to be append, not replace. That's
probably what Grant was alluding to before.
Prepend, not append, I believe., e.g. here are the
You showed the message from me as:

Message-ID: <uk5ksr$e7da$***@dont-email.me>
References: <uk3d11$3vnan$***@dont-email.me> <ikn9N.30033$rx%***@fx47.iad>

The one from you just now is:

Message-ID: <LYs9N.153256$***@fx17.iad>
References: <uk3d11$3vnan$***@dont-email.me> <ikn9N.30033$rx%***@fx47.iad> ...
<uk5ksr$e7da$***@dont-email.me>

ie the uk5 message-id that you got from me has clearly been
APPENDED to the references that you received from me.

And that is in accordance with what the spec says:

https://www.w3.org/Protocols/rfc850/rfc850.html

If the original article does have a References line, the
followup article should have a References line containing
the text of the original References line, a blank, and
the message ID of the original article.

I don't think it is reasonable for small systems
to support infinitely-long lines though. I'll probably
violate the spec and just have a single reference.
Post by Scott Lurndal
Post by Paul Edwards
https://www.w3.org/Protocols/rfc977/rfc977.html
didn't have it.
try 3977.
Nope, "References" isn't documented there either.

What's wrong with the reference (850) I already gave?

BFN. Paul.
wolfgang kern
2023-11-29 03:03:35 UTC
Permalink
On 29/11/2023 01:52, Paul Edwards wrote:
...
Post by Paul Edwards
What's wrong with the reference (850) I already gave?
w/o the complete reference path you mess up our threads by showing
several topics with the same RE:wording.
stop playing if you don't want to become reported to ES and google.
use any test-group with your use-net attempts until ready.
__
wolfgang
Grant Taylor
2023-11-29 05:00:26 UTC
Permalink
Post by wolfgang kern
w/o the complete reference path you mess up our threads by showing
several topics with the same RE:wording.
That's contrary to my understanding.

References MUST reflect the Message-ID: of the message you are replying to.

References SHOULD also reflect any parent Message-ID:s too.

N.B. MUST vs SHOULD

N.B. There is an effective upper limit on the (logical) line length of
the References: header. As such, once you get far enough in, the older
Message-ID:s fall off the end.
--
Grant. . . .
wolfgang kern
2023-11-29 06:31:16 UTC
Permalink
Post by Grant Taylor
Post by wolfgang kern
w/o the complete reference path you mess up our threads by showing
several topics with the same RE:wording.
That's contrary to my understanding.
References MUST reflect the Message-ID: of the message you are replying to.
References SHOULD also reflect any parent Message-ID:s too.
N.B. MUST vs SHOULD
N.B. There is an effective upper limit on the (logical) line length of
the References: header.  As such, once you get far enough in, the older
Message-ID:s fall off the end.
yeah, I meant _complete_ in terms of not missing what's required.
I didn't say "all of it".
__
wolfgang
Paul Edwards
2023-11-30 21:06:21 UTC
Permalink
Post by wolfgang kern
Post by Paul Edwards
What's wrong with the reference (850) I already gave?
w/o the complete reference path
1. You quoted a reference to an RFC. There is nothing
wrong with the RFC 850 I gave.

2. As you have already conceded, the complete reference
is not required (apparently - although I didn't see that
in the old RFC). Just the most recent message-id. BTW,
I just found out that the References line isn't one huge
long line. There are two per line, but presumably that is
flexible. So the References going out in this message is
going to be:

References: <uk3d11$3vnan$***@dont-email.me> <ikn9N.30033$rx%***@fx47.iad>
<uk5ksr$e7da$***@dont-email.me> <LYs9N.153256$***@fx17.iad>
<uk6200$g1fe$***@dont-email.me> <uk69m9$kt41$***@dont-email.me>

I think I previously sent one with 4 in one line,
which may have exceeded some limit, but I was
replying to one with 3 in one line that exceeded
80 characters itself.

3. All my messages (including the one you replied to)
have the complete References already. I haven't once
sent out a partial References. I previously
wasn't sending out ANY References, but I fixed that as soon
as someone told me what the technical issue was.
Post by wolfgang kern
you mess up our threads by showing
several topics with the same RE:wording.
That may happen on your news reader, but it doesn't
happen on Google Groups, so I wasn't aware there was
a problem until someone alerted me, and then I fixed
it as soon as someone told me what I was doing wrong.
Post by wolfgang kern
stop playing
I'm not "playing". I have always posted what I believed
were technically accurate messages.
Post by wolfgang kern
if you don't want to become reported to ES and google.
I don't see how I can be reported to google when I'm
not posting anything via google (because I can't anymore).

And although there is nothing preventing someone from
reporting me to ES, I think that is a very lousy thing
to do - trying to get me (potentially) banned because of
a threading issue. Also it would be a pretty harsh
judgement - this is an "alt" group. There are no rules
and no moderator. That's why Rick was able to post all
that religious stuff. Your only choice was to block him
at an individual level. In this case, you're assuming
that there is a rule that I can't start a new thread
with the same subject, and that ES will agree. See above
about "no rules". But I agree they may decide it's a
rule/non-rule violation regardless.
Post by wolfgang kern
use any test-group with your use-net attempts until ready.
I already did exactly that. I posted to an eternal september
test group. I was able to retrieve that message. It looked
fine to me (ES added a stack of header lines to my original).
Then I made a new post in alt.os.development.
That then appeared in google groups (which I can still read),
and also looked fine.

It was only later when I was replying to messages that an
issue (not visible to me via google groups) was discovered
in my software/procedure, and it was fixed in hours.

And you're already ready to report me to the authorities
as your initial "go to" solution? My goodness.
Grant Taylor
2023-12-01 00:55:16 UTC
Permalink
Post by Paul Edwards
I just found out that the References line isn't one huge
long line.
Not necessarily.
Post by Paul Edwards
There are two per line, but presumably that is
flexible. So the References going out in this message is
From memory -- I'm not finding and reading an RFC for this -- there are
basically three lengths that come into play.

- what will fit in a ""standard window, generally considered to be
72-78 characters.
- a physical line length
- a logical line length

I believe that the physical line length (for email, which news articles
have significant overlap with) is 1000 characters, including trailing
<CR><LF> so effectively 998 characters on the line.

I believe that theoretically the logical, unfolded / unwrapped line is
unlimited. However going beyond the physical line length of ~1k tends
to tickle bugs in things like there is no tomorrow.

Many things use a physical line length of < 80 characters and a logical
line length of < 1000 characters.

I say this to say that you could put all references and the header on
one line as long as it's less than the physical line length, even if
it's longer than the ~80 characters people consider to be good netiquette.
Post by Paul Edwards
I think I previously sent one with 4 in one line,
which may have exceeded some limit, but I was
replying to one with 3 in one line that exceeded
80 characters itself.
Four entries and a References header should have hardly scratched the
physical line length unless they were obscenely long Message-IDs. That
being said, they may have crossed the netiquette suggested 80 characters.

N.B. the 80 characters goes back to fixed with terminals and is IMHO
outdated. I now use terminals at least 132 characters wide and
frequently widen them. But the idea of the netiquette is to avoid
inconveniencing others and be considerate of their terminals by making
small changes in what you do.
Post by Paul Edwards
That may happen on your news reader, but it doesn't
happen on Google Groups, so I wasn't aware there was
a problem until someone alerted me, and then I fixed
it as soon as someone told me what I was doing wrong.
There is another form of threading contrary to the References: /
In-Reply-To: header and that's based on clients trying to be smart and
guess based on the Subject: header and date and maybe the opening quote
in the body. This is pseudo threading that works part of the time and
fails miserably other parts of the time.
Post by Paul Edwards
I'm not "playing". I have always posted what I believed
were technically accurate messages.
I trust that you are sending messages in good faith. However there is
one small nitpick that may be considered playing by some. (More below.)
Post by Paul Edwards
I don't see how I can be reported to google when I'm
not posting anything via google (because I can't anymore).
The only thing that I see that might be not 100% pure / innocent is
testing things about message threading in the alt.os.development
newsgroup. Usenet message threading may not be on topic in /this/
newsgroup.

That being said, so what.

Your /off/ /topic/ content became off topic as part of a reply / comment
/ complaint to an on topic post. You carried the conversation forward
in situ.

I'm continuing to carry your conversation forward in situ as well.

We are having polite, civil, productive conversation. Even if it's not
100% on topic for /this/ /newsgroup/.

Other than that, I don't see anything to complain about.
Post by Paul Edwards
And although there is nothing preventing someone from
reporting me to ES, I think that is a very lousy thing
to do - trying to get me (potentially) banned because of
a threading issue. Also it would be a pretty harsh
judgement - this is an "alt" group. There are no rules
and no moderator.
I'd be flabbergasted if you got anything more than a "please stay on
topic in $NEWSGROUP" type comment. Maybe if you repeatedly did it and
had multiple complaints and multiple such comments. But then, it would
not be a surprise to you.
Post by Paul Edwards
That's why Rick was able to post all
that religious stuff. Your only choice was to block him
at an individual level. In this case, you're assuming
that there is a rule that I can't start a new thread
with the same subject, and that ES will agree. See above
about "no rules". But I agree they may decide it's a
rule/non-rule violation regardless.
I also suspect that the E-S newsmaster has their hands full dealing with
actual spam from Google Groups.

I suspect that they would take one look at this and either ask "why am I
being bothered with this" or send you a message saying "please try to
stay on topic".

If anything, the comment about reporting you is more likely meant to
intimidate you.
Post by Paul Edwards
I already did exactly that. I posted to an eternal september
test group. I was able to retrieve that message. It looked
fine to me (ES added a stack of header lines to my original).
Then I made a new post in alt.os.development.
That then appeared in google groups (which I can still read),
and also looked fine.
It was only later when I was replying to messages that an
issue (not visible to me via google groups) was discovered
in my software/procedure, and it was fixed in hours.
I believe that Google Groups supports subject based threading.
Post by Paul Edwards
And you're already ready to report me to the authorities
as your initial "go to" solution? My goodness.
Chuckle.
--
Grant. . . .
wolfgang kern
2023-12-01 06:30:50 UTC
Permalink
On 30/11/2023 22:06, Paul Edwards wrote:
...
Post by Paul Edwards
And you're already ready to report me to the authorities
as your initial "go to" solution? My goodness.
not me of course, it was just a brief warning.
__
wolfgang

Loading...