I've been hunting down an incoherent case in Gnus while downloading mail
from a Gmail POP3 server. Today I spotted the following case. In the
POP3, I found 8 messages:

+OK 8 messages (80988 bytes)
1 20295
2 4583
3 20295
4 20295
5 2725
6 5487
7 2725
8 4583
..

When Gnus downloaded all the main, it found the 8 messages, but the
amount of bytes was much greater. (The extra logging has been added by
me.) Notice how the total-size matches the advertised total-size by the
POP3 server.

Checking new news...
Reading active file via nnml...
Reading incoming mail from pop...
size 0, total-size 80988
pop3 retrieved 0KB (0%)
size 237, total-size 80988
pop3 retrieved 0KB (0%)
size 751, total-size 80988
pop3 retrieved 0KB (0%)
size 1777, total-size 80988
[...]
pop3 retrieved 80KB (99%)
size 81659, total-size 80988
pop3 retrieved 81KB (100%)
size 81865, total-size 80988
pop3 retrieved 81KB (101%)
size 83255, total-size 80988
pop3 retrieved 83KB (102%)
size 84645, total-size 80988
[...]
size 511138, total-size 80988
pop3 retrieved 511KB (631%)
Wrote c:/sys/emacs/usr/Mail/mail/misc/1799
Wrote c:/sys/emacs/usr/Mail/mail/misc/1800
Wrote c:/sys/emacs/usr/Mail/mail/misc/1801
Wrote c:/sys/emacs/usr/Mail/mail/misc/1802
Wrote c:/sys/emacs/usr/Mail/mail/misc/1803
Wrote c:/sys/emacs/usr/Mail/mail/misc/1804
Wrote c:/sys/emacs/usr/Mail/mail/misc/1805
Wrote c:/sys/emacs/usr/Mail/mail/misc/1806
nnml: Reading incoming mail (8 new)...done
--8<-------------------------------------------------------->8---

On disk, almost all messages have similar sizes to the sizes advertised
by the POP3 server, except for message 1799. I got a couple of PDF
files in one of these 8 messages and so it looks like the POP3 somehow
did not report the size of this message as I expected. Is there any
chance the POP3 is actually correct in doing what it's doing?

--8<-------------------------------------------------------->8---
%ls -l | awk '$8 >= 1799 && $8 <= 1806 { total += $5; print; } \
END { print(total) }'
-rw-rw-rw- 1 x None 446971 2024-05-21 17:43 1799
-rw-rw-rw- 1 x None 8441 2024-05-21 17:43 1800
-rw-rw-rw- 1 x None 16925 2024-05-21 17:43 1801
-rw-rw-rw- 1 x None 20494 2024-05-21 17:43 1802
-rw-rw-rw- 1 x None 6513 2024-05-21 17:43 1803
-rw-rw-rw- 1 x None 5688 2024-05-21 17:43 1804
-rw-rw-rw- 1 x None 2925 2024-05-21 17:43 1805
-rw-rw-rw- 1 x None 4781 2024-05-21 17:43 1806
512738
--8<-------------------------------------------------------->8---

(*) The attachments

I'm showing just the parts of message 1799 that show there were two
attachments. Given just the size of this message, it seems the POP3
really is wrong.

--000000000000eda27d0618f68db7
Content-Type: multipart/alternative; boundary="000000000000eda2780618f68db5"

--000000000000eda2780618f68db5
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

[...]

--000000000000eda2780618f68db5
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

[...]

--000000000000eda2780618f68db5--
--000000000000eda27d0618f68db7
Content-Type: application/pdf;
name="[...]"
Content-Disposition: attachment;
filename="[...]"
Content-Transfer-Encoding: base64
Content-ID: <f_lwgexh0q1>
X-Attachment-Id: f_lwgexh0q1

JVBERi0xLjQKJdPr6eEKMSAwIG9iago8PC9DcmVhdG9yIChNb3ppbGxhLzUuMCBcKFdpbmRvd3Mg
TlQgMTAuMDsgV2luNjQ7IHg2NFwpIEFwcGxlV2ViS2l0LzUzNy4zNiBcKEtIVE1MLCBsaWtlIEdl
[...]
Ci9JbmZvIDEgMCBSPj4Kc3RhcnR4cmVmCjEyMjk0NgolJUVPRg==
--000000000000eda27d0618f68db7
Content-Type: application/pdf;
name="[...].pdf"
Content-Disposition: attachment;
filename="[...].pdf"
Content-Transfer-Encoding: base64
Content-ID: <f_lwgexh090>
X-Attachment-Id: f_lwgexh090

JVBERi0xLjQKJdPr6eEKMSAwIG9iago8PC9UaXRsZSAoU0FQSUVOUykKL0NyZWF0b3IgKE1vemls
bGEvNS4wIFwoV2luZG93cyBOVCAxMC4wOyBXaW42NDsgeDY0XCkgQXBwbGVXZWJLaXQvNTM3LjM2
[...]
MDQgMCBSCi9JbmZvIDEgMCBSPj4Kc3RhcnR4cmVmCjE4NjUyNgolJUVPRgo=
--000000000000eda27d0618f68db7--
--8<-------------------------------------------------------->8---