Re: vger not relaying some of Junio's messages today?

[Johannes Schindelin]

Hi Brandon,

On Mon, 8 May 2017, Brandon Williams wrote:

> On 05/08, Johannes Schindelin wrote:
> 
> > On Sat, 6 May 2017, Ævar Arnfjörð Bjarmason wrote:
> > 
> > > I have one [script] to git am a patch from a msgid, thought I should
> > > write something to handle a series in some DWIM fashion (e.g. apply
> > > the latest continuous sequence of patches matching --author) but
> > > figured that someone's probably wrote this already & I don't need to
> > > hack it up myself...
> > 
> > You probably missed my previous mails mentioning
> > 
> > https://github.com/git-for-windows/build-extra/blob/master/apply-from-public-inbox.sh
> > 
> > You can use this script to apply single patches (identified by their
> > Message-ID), and patch series (identified by their cover letter's
> > Message-ID).
> > 
> > As I mentioned at the Contributors' Summit at GitMerge 2017: I would
> > *love* to collaborate on tools that make any part of the
> > contribution/review process less cumbersome than it is right now.
> 
> Yeah its not the most streamlined process.  I'm sure everyone writes
> their own scripts (like I did) tailored to their workflow.

I am sure you are right. What a waste of time, for everybody to come up
with essentially the same sort of scripts, just to be able to participate.

> For example I just tag a bunch of mails in mutt and then have a scripts
> which 'git am's them on a branch/base of my choosing.  But its specific
> to my workflow so idk how useful it would be to others :(

Hmm. So it looks more and more as if you *have* to use mutt in order to
be rewarded with the option for an efficient workflow.

I'm just so used to my good ole' Alpine. And others may be so used to
their Thunderbird, Outlook, GMail, whatevs.

But hey, maybe the vger woes will eventually become so bad that even mutt
and NNTP users will be affected negatively. At that point, we may look
into alternatives.

Ciao,
Dscho

Re: vger not relaying some of Junio's messages today?

[Brandon Williams]

On 05/07, Eric Wong wrote:
> Samuel Lijin  wrote:
> > > Samuel Lijin  wrote:
> > >> Yep, I see these on public-inbox.org/git/ but not in my gmail inbox:
> > >
> > > Hi Samuel, check your Spam box (and move it to a normal inbox so
> > > they can train it).  Gmail filters are known to trigger happy
> > > and incorrectly flag messages.  It's been a problem on LKML,
> > > too.
> > 
> > Sorry, should've been clearer - I did check my spambox in my original
> > message. Some old patches from Brandon were in there, but the ones I
> > mentioned in my original message just seem to have been dropped.
> 
> Yikes.  I wonder if Gmail automatically nukes messages if
> they end up in enough Spam boxes...  Really hoping Googlers
> can do something about this (or better, more people start
> running their own SMTP servers, again).

That's quite odd, none of my coworkers are seeing behavior like this (my
mail going to spam).

-- 
Brandon Williams

Re: vger not relaying some of Junio's messages today?

[Brandon Williams]

On 05/08, Johannes Schindelin wrote:
> Hi Ævar,
> 
> On Sat, 6 May 2017, Ævar Arnfjörð Bjarmason wrote:
> 
> > I have one [script] to git am a patch from a msgid, thought I should
> > write something to handle a series in some DWIM fashion (e.g. apply the
> > latest continuous sequence of patches matching --author) but figured
> > that someone's probably wrote this already & I don't need to hack it up
> > myself...
> 
> You probably missed my previous mails mentioning
> 
> https://github.com/git-for-windows/build-extra/blob/master/apply-from-public-inbox.sh
> 
> You can use this script to apply single patches (identified by their
> Message-ID), and patch series (identified by their cover letter's
> Message-ID).
> 
> As I mentioned at the Contributors' Summit at GitMerge 2017: I would
> *love* to collaborate on tools that make any part of the
> contribution/review process less cumbersome than it is right now.

Yeah its not the most streamlined process.  I'm sure everyone writes
their own scripts (like I did) tailored to their workflow.  For example
I just tag a bunch of mails in mutt and then have a scripts which 'git
am's them on a branch/base of my choosing.  But its specific to
my workflow so idk how useful it would be to others :(


-- 
Brandon Williams

Re: vger not relaying some of Junio's messages today?

[Johannes Schindelin]

Hi Ævar,

On Sat, 6 May 2017, Ævar Arnfjörð Bjarmason wrote:

> I have one [script] to git am a patch from a msgid, thought I should
> write something to handle a series in some DWIM fashion (e.g. apply the
> latest continuous sequence of patches matching --author) but figured
> that someone's probably wrote this already & I don't need to hack it up
> myself...

You probably missed my previous mails mentioning

https://github.com/git-for-windows/build-extra/blob/master/apply-from-public-inbox.sh

You can use this script to apply single patches (identified by their
Message-ID), and patch series (identified by their cover letter's
Message-ID).

As I mentioned at the Contributors' Summit at GitMerge 2017: I would
*love* to collaborate on tools that make any part of the
contribution/review process less cumbersome than it is right now.

Ciao,
Dscho

Re: vger not relaying some of Junio's messages today?

[Junio C Hamano]

Eric Wong  writes:

> Samuel Lijin  wrote:
>> Sorry, should've been clearer - I did check my spambox in my original
>> message. Some old patches from Brandon were in there, but the ones I
>> mentioned in my original message just seem to have been dropped.
>
> Apparently, vger also throttles mail to gmail aggressively, so
> maybe some show up eventually:
>
> https://marc.info/?i=20170501.105057.824365162373797902.da...@davemloft.net
>

Thanks for a pointer. That does explain the symptom I saw number of
times, seeing list traffic that directly came to me hours before I
see a copy on the public list mirrors like gmane and public-inbox.

As it is a pain to access gmail inbox via imap (I was told that
something called "offline imap" may alleviate the pain, but I
haven't tried it yet), I tend to use NNTP interface to either gmane
or public-inbox almost exclusively, so like everybody else, it often
takes a message hours before I see it due to this throttling.  I do
not see the message I am responding to on public-inbox, and I am
responding to the copy I got direct from you, for example.

Re: vger not relaying some of Junio's messages today?

[Eric Wong]

Samuel Lijin  wrote:
> > Samuel Lijin  wrote:
> >> Yep, I see these on public-inbox.org/git/ but not in my gmail inbox:
> >
> > Hi Samuel, check your Spam box (and move it to a normal inbox so
> > they can train it).  Gmail filters are known to trigger happy
> > and incorrectly flag messages.  It's been a problem on LKML,
> > too.
> 
> Sorry, should've been clearer - I did check my spambox in my original
> message. Some old patches from Brandon were in there, but the ones I
> mentioned in my original message just seem to have been dropped.

Yikes.  I wonder if Gmail automatically nukes messages if
they end up in enough Spam boxes...  Really hoping Googlers
can do something about this (or better, more people start
running their own SMTP servers, again).

Re: vger not relaying some of Junio's messages today?

[Eric Wong]

Samuel Lijin  wrote:
> Sorry, should've been clearer - I did check my spambox in my original
> message. Some old patches from Brandon were in there, but the ones I
> mentioned in my original message just seem to have been dropped.

Apparently, vger also throttles mail to gmail aggressively, so
maybe some show up eventually:

https://marc.info/?i=20170501.105057.824365162373797902.da...@davemloft.net



On a side note, I just started trying the "mutt-patched"
package in Debian stable.  It looks like "mutt" package will
absorb it's functionality in future releases, but maybe more
mutt users can have NNTP support in their favorite MUA :)

Re: vger not relaying some of Junio's messages today?

[Eric Wong]

Eric Wong  wrote:
> (OTOH, I noticed a thread/mbox download bug in public-inbox,
> https://public-inbox.org/git/xmqqmvaq702u@gitster.mtv.corp.google.com/t.mbox.gz
> only shows two messages out of many.   Will need to fix that...)

I think I fixed that bug, at least:
https://public-inbox.org/meta/20170507110328.GA20102@whir/

And vger is still down...

Re: vger not relaying some of Junio's messages today?

[Eric Wong]

Junio C Hamano  wrote:
> As it is a pain to access gmail inbox via imap (I was told that
> something called "offline imap" may alleviate the pain, but I
> haven't tried it yet),

offlineimap isn't bad; I've been using it since 2003-2004
and can say it's easier-to-setup and more reliable than
isync/mbsync, especially with not-so-great "IMAP" servers.

offlineimap also has specific examples in the docs for Gmail's
weird IMAP interface.  My main complaints about it are high memory
usage and slowness.

> I tend to use NNTP interface to either gmane
> or public-inbox almost exclusively, so like everybody else, it often
> takes a message hours before I see it due to this throttling.  I do
> not see the message I am responding to on public-inbox, and I am
> responding to the copy I got direct from you, for example.

Looks like vger is down, my mail server can't make connections
to it on port 25 at all.  Good thing we have a reply-to-all
convention here :)

Re: vger not relaying some of Junio's messages today?

[Eric Wong]

Ævar Arnfjörð Bjarmason  wrote:
> Thanks a lot for public-inbox, my only problem with it is that it
> doesn't cover every single mailing list I'm on, just git :)

Yes, I'm hoping more folks can start running their own
instances.  I notice you're on p5p where I started to
occasionally lurk a few years back.  Is that something you'd be
interesting in mirroring? :)

Some lists I'm involved in tend to be centralized
(reply-to-list, not reply-to-all); so UI changes will be needed
in public-inbox...

> Are you or someone else maintaining some ancillary scripts for it? I
> probably need to fix my patch workflow but my usual mode is browsing
> in GMail & then manually 'git am'-ing some file I find with git-log
> commands.

Not really.  I started out ssoma(*) just for archival/mirroring,
but it's mostly forgotten at this point.  I would rather people
build on existing protocols like NNTP.

> I have one to git am a patch from a msgid, thought I should write
> something to handle a series in some DWIM fashion (e.g. apply the
> latest continuous sequence of patches matching --author) but figured
> that someone's probably wrote this already & I don't need to hack it
> up myself...

I've been meaning to wire up the web search to allow downloading
an entire mbox worth of search results.  But it could be a bit
expensive for some queries and there needs to be a limit.  The
current limit for web display is too small; so maybe 1000
messages per-mbox at a time to avoid DoS-ing the server.

(*) some-sort-of-mail archiver: git clone git://80x24.org/ssoma

Re: vger not relaying some of Junio's messages today?

[Samuel Lijin]

On Sat, May 6, 2017 at 3:50 PM, Eric Wong  wrote:
> (I have no idea what Jonathan Tirado wrote; it was encrypted (but
>  sent to a public list).
>
> Samuel Lijin  wrote:
>> Yep, I see these on public-inbox.org/git/ but not in my gmail inbox:
>
> Hi Samuel, check your Spam box (and move it to a normal inbox so
> they can train it).  Gmail filters are known to trigger happy
> and incorrectly flag messages.  It's been a problem on LKML,
> too.

Sorry, should've been clearer - I did check my spambox in my original
message. Some old patches from Brandon were in there, but the ones I
mentioned in my original message just seem to have been dropped.

Re: vger not relaying some of Junio's messages today?

[Ævar Arnfjörð Bjarmason]

On Sat, May 6, 2017 at 10:50 PM, Eric Wong  wrote:
> (I have no idea what Jonathan Tirado wrote; it was encrypted (but
>  sent to a public list).
>
> Samuel Lijin  wrote:
>> Yep, I see these on public-inbox.org/git/ but not in my gmail inbox:
>
> Hi Samuel, check your Spam box (and move it to a normal inbox so
> they can train it).  Gmail filters are known to trigger happy
> and incorrectly flag messages.  It's been a problem on LKML,
> too.
>
>> - Brandon [RFC 01/14] through [RFC 14/14] convert dir.c to take an
>> index parameter
>
> Ironically, Brandon is a Google employee and Gmail seems to not
> like his messages.  The only flag I see from SpamAssassin in
> public-inbox is HEADER_FROM_DIFFERENT_DOMAINS which happens to
> every message because it's relayed via kernel.org
>
> Maybe Brandon can escalate this internally in Google...
>
> (OTOH, I noticed a thread/mbox download bug in public-inbox,
> https://public-inbox.org/git/xmqqmvaq702u@gitster.mtv.corp.google.com/t.mbox.gz
> only shows two messages out of many.   Will need to fix that...)
>
>> - Johanne's Coverity patch series
>
> Likewise, but he also uses freemail domain (nothing wrong with
> that, but it raises one flag on SA).  I noticed at least one
> of his did trigger the RCVD_IN_SORBS_SPAM in SpamAssassin, but that's
> a common problem with all freemail providers (including Gmail).
>
>
> Anyways, I'm glad the SpamAssassin seems to be working well on
> public-inbox and would be grateful to know if there's false
> positives and missing messages.
>
> One of the major reasons I started public-inbox was because I
> lack faith in my ability to run my little list server (with
> mlmmj or mailman) and be able to successfully relay to the big
> providers.  Anyways, I'm glad it's helping readers of git@vger,
> too :)

Thanks a lot for public-inbox, my only problem with it is that it
doesn't cover every single mailing list I'm on, just git :)

Are you or someone else maintaining some ancillary scripts for it? I
probably need to fix my patch workflow but my usual mode is browsing
in GMail & then manually 'git am'-ing some file I find with git-log
commands.

I have one to git am a patch from a msgid, thought I should write
something to handle a series in some DWIM fashion (e.g. apply the
latest continuous sequence of patches matching --author) but figured
that someone's probably wrote this already & I don't need to hack it
up myself...

Re: vger not relaying some of Junio's messages today?

[Eric Wong]

(I have no idea what Jonathan Tirado wrote; it was encrypted (but
 sent to a public list).

Samuel Lijin  wrote:
> Yep, I see these on public-inbox.org/git/ but not in my gmail inbox:

Hi Samuel, check your Spam box (and move it to a normal inbox so
they can train it).  Gmail filters are known to trigger happy
and incorrectly flag messages.  It's been a problem on LKML,
too.

> - Brandon [RFC 01/14] through [RFC 14/14] convert dir.c to take an
> index parameter

Ironically, Brandon is a Google employee and Gmail seems to not
like his messages.  The only flag I see from SpamAssassin in
public-inbox is HEADER_FROM_DIFFERENT_DOMAINS which happens to
every message because it's relayed via kernel.org

Maybe Brandon can escalate this internally in Google...

(OTOH, I noticed a thread/mbox download bug in public-inbox,
https://public-inbox.org/git/xmqqmvaq702u@gitster.mtv.corp.google.com/t.mbox.gz
only shows two messages out of many.   Will need to fix that...)

> - Johanne's Coverity patch series

Likewise, but he also uses freemail domain (nothing wrong with
that, but it raises one flag on SA).  I noticed at least one
of his did trigger the RCVD_IN_SORBS_SPAM in SpamAssassin, but that's
a common problem with all freemail providers (including Gmail).


Anyways, I'm glad the SpamAssassin seems to be working well on
public-inbox and would be grateful to know if there's false
positives and missing messages.

One of the major reasons I started public-inbox was because I
lack faith in my ability to run my little list server (with
mlmmj or mailman) and be able to successfully relay to the big
providers.  Anyways, I'm glad it's helping readers of git@vger,
too :)

Re: vger not relaying some of Junio's messages today?

[Samuel Lijin]

Yep, I see these on public-inbox.org/git/ but not in my gmail inbox:

- Brandon [RFC 01/14] through [RFC 14/14] convert dir.c to take an
index parameter
- Johanne's Coverity patch series

Re: vger not relaying some of Junio's messages today?

[Jonathan Tirado]

binnbmf6RUfBU.bin
Description: application/pgp-encrypted


encrypted.asc
Description: Binary data

Re: vger not relaying some of Junio's messages today?

[Junio C Hamano]

Jeff King  writes:

> On Fri, Jan 27, 2017 at 03:57:53AM +, Eric Wong wrote:
>
>> I noticed both of these are are missing from my archives
>> (which rejects messages unless they come from vger):
>> 
>> 
>> 
>
> I don't have them either, so presumably vger ate them (I usually delete
> my cc copies after reading and keep the list ones, and I now have
> neither).
>
>> Not sure if there's something up with vger or Junio's setup because
>> other messages (even from Junio) are getting through fine.
>
> Sporadic failures, especially on a single topic, imply that something in
> the content may triggered the taboo filter (though often that takes out
> replies, too, which this doesn't seem to have done).

I think I figured this out yesterday.  

It was a dot somewhere in the name of a recipient on To:/Cc: line
that is NOT double-quoted.  For this thread, brian's human-readable
name "brian m. carlson" was double-quoted in the original I
responded to and it also was double-quoted in my response.
But there was another recipient with a dot after the middle initial,
which was NOT double-quoted in the original.  

My response seems to have been eaten because that name was not
double-quoted.  I saw that happen yesterday in another thread with
the same recipient, and resent with the only change to double-quote
that name and it went through.

Re: vger not relaying some of Junio's messages today?

[Jeff King]

On Fri, Jan 27, 2017 at 03:57:53AM +, Eric Wong wrote:

> I noticed both of these are are missing from my archives
> (which rejects messages unless they come from vger):
> 
> 
> 

I don't have them either, so presumably vger ate them (I usually delete
my cc copies after reading and keep the list ones, and I now have
neither).

> Not sure if there's something up with vger or Junio's setup because
> other messages (even from Junio) are getting through fine.

Sporadic failures, especially on a single topic, imply that something in
the content may triggered the taboo filter (though often that takes out
replies, too, which this doesn't seem to have done).

-Peff