[9fans] Gmail and upas
David Welling
Plan 9. I've been using it for about a week now, and things are coming
along well.
I'm having a bit of an issue, and it may entirely stem from my lack of
familiarity on the system, so I apologize in advance if this turns out
to be a silly problem:
When I throw
upas/fs -f /imaps/imap.gmail.com/sidi...@gmail.com
into my profile, (after the plumber, before I start rio; as per the
suggestion regarding mail on the wiki), I get a ludicrous amount of
errors after a while akin to this:
download 1768: i/o error: i/o on hungup channel
download 1769: i/o error: i/o on hungup channel
download 1770: i/o error: i/o on hungup channel
download 1771: i/o error: i/o on hungup channel
Et cetera. I did have copious amounts of mail in this box, somewhere
in the neighborhood of 3500. When I first ran this procedure, it
worked pretty well, albeit it slowly (3500 is a lot of messages to go
through). I got some similar messages, but I was able to access my
mail, and my machine seemed to run just fine. I used gmail's web
interface to clean out about 2000 of those mails, and rebooted the
system. Now, with these errors, the system grinds to a halt and there
is no responsiveness to speak of.
If any of you folks have any constructive suggestions, I certainly
appreciate the input. I perused various manpages and tried to dig
through the archives here, but haven't seen a problem quite like mine.
Thanks in advance,
Sid
lu...@proxima.alt.za
> in the neighborhood of 3500.
Whenever I allow my IMAP mailbox to grow too large, upas also grinds
to a halt. I haven't yet done justice to Erik Quanstrom's nupas to
see if it makes a difference because it is too big a leap at this
point in time, but you may want to put it to the test. It's in
/n/sources/contrib/quanstro/src/nupas if memory serves.
Be warned that it is different in the way it represents the folder, so
the end results may be quite dramatic.
++L
erik quanstrom
> the end results may be quite dramatic.
? the file system interface is not changed.
- erik
Robert Raschke
actually implement IMAP to the standard. There are quite a few odd
behaviours. A notable one has to do with deletion of emails, but I
can't find the exact reference just now. (I can try digging if you're
really intereested.)
Robby
erik quanstrom
thumbprint to /sys/lib/tls/mail:
x509 sha1=e221be6be22afd3b3244199476cbb136da4ad02d cn=*.gmail.com
unless you've added this, the messages are downright cryptic.
> into my profile, (after the plumber, before I start rio; as per the
> suggestion regarding mail on the wiki), I get a ludicrous amount of
> errors after a while akin to this:
>
> download 1768: i/o error: i/o on hungup channel
> download 1769: i/o error: i/o on hungup channel
> download 1770: i/o error: i/o on hungup channel
> download 1771: i/o error: i/o on hungup channel
i don't get these messages but they seem plausable
for a missing thumbprint. the thumbprint is required;
there's no way to turn off thumbprint checking.
> Et cetera. I did have copious amounts of mail in this box, somewhere
> in the neighborhood of 3500. When I first ran this procedure, it
> worked pretty well, albeit it slowly (3500 is a lot of messages to go
> through). I got some similar messages, but I was able to access my
while imap4 support with nupas has some warts, nupas
has users with local mailboxes that have 8500 messages.
so your gmail box seems doable.
the source for nupas is at /n/sources/contrib/quanstro/nupas.
but if you want to use it with gmail imap, you may wish to
wait a few weeks or so. i'll make an announcement.
by the way, i don't know of any way that gmail imap is not standard.
perhaps this is in some esoteric corners of the protocol i'm not
familar with. for the basic stuff, i haven't seen any issues at all.
- erik
Rudolf Sykora
seems you use IMAP to read gmail. I usually read my gmail mail through
my web browser, which is not a problem from opera/firefox in linux.
However, I can't do the same from plan9. Neither abaco, nor charon
work. Is this so for everyone or just for me? Thanks.
Ruda
Robert Raschke
> by the way, i don't know of any way that gmail imap is not standard.
> perhaps this is in some esoteric corners of the protocol i'm not
> familar with. for the basic stuff, i haven't seen any issues at all.
The two immediate issues I am aware of (via the UW-IMAP list) is that
MIME part sizes may be reported wrongly and that setting \Deleted on a
message will lead to it getting silently removed through an
auto-expunge that gmail does completely on its own (I believe this
latter behaviour can now be turned off through an experimental
setting).
Robby
erik quanstrom
> > perhaps this is in some esoteric corners of the protocol i'm not
> > familar with. for the basic stuff, i haven't seen any issues at all.
>
> The two immediate issues I am aware of (via the UW-IMAP list) is that
> MIME part sizes may be reported wrongly and that setting \Deleted on a
> message will lead to it getting silently removed through an
> auto-expunge that gmail does completely on its own (I believe this
> latter behaviour can now be turned off through an experimental
> setting).
calling that "non-standard" is mightly charitable. if that were my software,
folk would be calling those bugs.
- erik
Eris Discordia
GMail's normal web interface is a heavy web application relying on AJAX.
That is unavailable on Plan 9's browsers from A (asynchronous), to J
(JavaScript), to X (XML). You may want to try to disable the advanced
interface, go with the basic HTML-only one, and then try it on Plan 9's
browsers. Even though, the login page alone could pose serious problems
with a basic unadorned browser like abaco.
People here seem to use IMAP to do their bidding but POP (POPS actually)
can be just fine and very simple if you only want to read your email and
you don't receive huge attachments that will clog your line if you even try
to read the text part of an a new email.
<http://plan9.escet.urjc.es/magic/man2html/4/upasfs> (try poptls)
FUN FACT: GMail works well with links/elinks.
--On Friday, November 21, 2008 3:43 PM +0100 Rudolf Sykora
a...@9srv.net
"out of the box". The failure isn't abaco, it's webfs. If you
take /n/sources/contrib/fgb/cookies.c, bind or copy it
over /sys/src/cmd/webfs/cookies.c, and rebuild, the new
webfs and abaco will work fine.
I don't know why fgb hasn't submitted it as a patch, and
I can't really evaluate the technical correctness of the
changes to cookie handling. But it works, and i've not seen
it break anything else.
David Welling
little more than 15 hours, I thank everyone for their help.
I have not yet tried nupas, but have gotten a reasonable solution working.
It appeared that while I was tweaking my setup (and rebooting often,
causing upas to re-download thousands of messages over and over again)
gmail decided I was a bot and gave me a SECTOR 4 LOCKDOWN, a temporary
status when an account is being used too much that prevents access.
With a little patience I rebooted my machine in the morning and went
to class. After class I came home to a brilliant display of faces and
acme-mail goodness.
So my solution, albeit temporary, causes long boot times (and prevents
me from having to reboot a ton (at least without first removing the
box from the interblarg). But, given that this machine is intended to
stay up for extended periods of time, I don't forsee that as an issue.
Maybe once I've really gotten a stronger feel for things I'll dedicate
a machine and just import /mail to my terminal.
Using pop3 access would eliminate this issue altogether, but I'm sure
all you plan9gmailers out there already know that gmail is a little
wonky with upas, due to gmail's nonconformity in forcing pop clients
into thinking that they are removing the server-side copy of the mail.
Oh well, I'll spend some more time tweaking and breaking. That's the
fun of it all.
Also, Ruda, you can get gmail's web interface working with abaco,
there's a patch you have to apply to webfs to get the cookies working
correctly. For more information peruse the plan9 wiki, someone threw
up a nice gmail page.
Iruata Souza
<eris.di...@gmail.com> wrote:
> FUN FACT: GMail works well with links/elinks.
>
does humans work well with that software?
iru
Eris Discordia
I do, but I doubt I qualify as human--from your point of view.
--On Friday, November 21, 2008 5:52 PM -0200 Iruata Souza
Iruata Souza
<eris.di...@gmail.com> wrote:
>> does humans work well with that software?
>
> I do, but I doubt I qualify as human--from your point of view.
>
since the software was written with you in mind, I suppose it fits well, then.
iru
ps: I love science
Eris Discordia
Yeah, and that must be UV.
P.S. I know I shouldn't be replying and further staining my already messed
up reputation here.
--On Friday, November 21, 2008 6:16 PM -0200 Iruata Souza
Federico G. Benavento
relax
http://groups.google.com/group/comp.os.plan9/msg/406fc491a206e562
http://abaco.oitobits.net/images/abaco-ss-00.jpg
--
Federico G. Benavento
Jack Johnson
I remember looking into the problem with charon in the past using my
feeble skills, so I just took another peek using:
charon -dbg dnop -docookies 1 -doscripts 1 -usessl v3 -starturl
'http://mail.google.com/mail/h' -dbgfile gmail.out
and based on the URL left in the address bar and the debug log, it
looks like it fails somewhere around the last continue parameter from
the last redirect. Authentication seems to be working; I can use
iGoogle but not Gmail.
There are some intermittent errors about SSL to
ssl.google-analytics.com, but I don't think these are critical.
One of the last requests seem to return a chunk of Javascript that
does browser detection:
https://mail.google.com/mail?view=page&name=browser&ver=1k96igf4806cy
I tried some of the tricks found around the net to turn off browser
detection with no success. I did find that the URL redirects seem to
behave differently if I don't enable Javascript, but then it fails
very strangely at the end with what appears to be an invalid request.
The last valid request is:
and then it follows up with:
GO TO https://www.google.com/accounts/'http:/mail.google.com/mail/h/19sso9tatmt7r/?ui=html&zy=l'
target frame name=
Startreq BS=12 for
https://www.google.com/accounts/'http:/mail.google.com/mail/h/19sso9tatmt7r/?ui=html&zy=l'
Chose NC=0 for BS 12, qlen=1
NC 0: starting runnetconn in connect state
http 0: dialing tcp!www.google.com!443
Waitreq
Waitpending
http 0: connected
warning: unknown header field: X-Content-Type-Options: nosniff
http 0: got response header:
HTTP/1.0 404 Not Found
which doesn't seem to match the continue parameter from the last request.
-J
Jack Johnson
> and then it follows up with:
>
> GO TO https://www.google.com/accounts/'http:/mail.google.com/mail/h/19sso9tatmt7r/?ui=html&zy=l'
>
I just disabled Javascript on Safari and tried
http:/mail.google.com/mail/h and eventually I get a URL like the above
with a *very* long auth parameter appended to the end, and then a usr
parameter with my email address.
-J
Eris Discordia
> http://abaco.oitobits.net/images/abaco-ss-00.jpg
Lucky for me, I did suggest that GMail basic may work with abaco and
encouraged Rudolf Sykora to try it. I just added one fun fact after it,
because GMail's damn heavy interface works on links/elinks and that needed
to be mentioned.
--On Saturday, November 22, 2008 1:10 AM -0200 "Federico G. Benavento"