Thus said Ken Hornstein on Wed, 06 Nov 2024 12:37:31 -0500:
> It frustrates me that people aren't complying with the obvious MTA
> limits, but here we are. And ... well, the simple fact of the matter
> is, as far as I can tell, everybody ELSE deals with huge lines just
> fine. I am forced to conclude that so should we.
I really would like to see if we can get UPS to recognize this. They are
big enough that it should be of concern to them. The problem is, nobody
rejects such emails. As Stephen rightly points out, it seems like only
something as big as Gmail could actually pull it off.
As I stated before, I completely agree with Ralph's assessment (I think
it was Ralph) that nmh should do nothing do address this. But that then
leaves the question, what should be done.
I went through my past emails and also discovered this same problem
with ups.com emails (I don't recall ups.com having this problem before
so it must be newer behavior). In fact, I find that their problem is
just as egregious as the one that I reported many years ago.
I wonder what the trend looks like. This awk command will print the
longest line of each nmh message file in this directory:
for f in $(grep -l -i '^From:.*ups.com' *); do printf '%s: ' $f; awk 'BEGIN {
l=0 } { t=length($0); if (t > l) l = t } END { print l }' $f; done | sort -k 1n
215: 677
243: 684
246: 626
1410: 661
1411: 661
1412: 661
1417: 668
1418: 668
1419: 668
1423: 487
1424: 487
1425: 487
1426: 53256
1427: 53256
1431: 53256
1721: 1427
1722: 1427
1723: 1427
1729: 1434
1730: 1434
1731: 1434
1736: 1145
1737: 1145
1738: 1145
1740: 53948
1741: 53948
1742: 53948
Message 215 is dated March 10 of 2023. Message 1462 (which is when it
seems that they started getting longer lines) is dated May 28 of 2024.
So it would seem that ups.com has only recently started sending emails
that violate the line lengths. Now is the time to alert them. The closer
to the time that they implemented whatever change, the better.
> I took a look at Andy's patch; I didn't go through ALL of the buffer
> handling logic just yet, but at first glance it seems reasonable.
I hope it's looked at fairly well. Long-established code shouldn't be
changed lightly in my opinion. Also, I won't be offended if someone
comes up with a better solution, or even if it's outright denied as it
has been.
Andy
