incubator-wave-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alain Levesque <albon...@wavewatchers.org>
Subject Re: Email bridge bot
Date Tue, 11 Jun 2013 16:22:20 GMT
I do have time also to be a ''regular user'' . Feel free to contact me as
neeeded. Bravo! Bruno


On Mon, Jun 10, 2013 at 9:56 PM, Alfredo Abambres <alfredoabambres@gmail.com
> wrote:

> Thank you Bruno for trying to make this. I can't be much of assistance on
> this point, but if you need a "regular user" to help you test it, just wave
> :-)
>
> http://alfredo.abambres.com
>
> *"Moving, always moving, and living inside movement". Rainer Maria Rilke*
>
>
> On Sat, Jun 8, 2013 at 1:03 AM, Bruno Gonzalez (aka stenyak) <
> stenyak@gmail.com> wrote:
>
> > Right now, in part due to its alpha state, and in part due to bugs (I
> can't
> > receive newBlip notifications, etc), emails are only sent when the user
> > writes "bot:send\n". At that very moment, the bot sends a single email.
> >
> > Regarding synchronization schedule, we could keep a list of "blips not
> yet
> > synced to email", each of which would have a timeout. Whenever the blip
> > contents is edited, the blip timeout gets reset. Blips that reach the
> > timeout command the bot to sync themselves. Having that basic mechanism,
> > there can be additional rules (for example, all ancestors of a blip have
> to
> > be synced before the child blip is synced. stuff like that).
> >
> > The timeout period could be configurable, and we can take existing
> > platforms are a reference. Some examples:
> >  - GMail's "undo" (the atrophied uncle of Wave's "edit") used to be
> > customizable from 0 to 30 seconds. Recently they increased the limit to
> 60
> > seconds.
> >  - Some forums and social networks allow to choose "inmediate" (zero
> > seconds) and "daily"/"weekly" (timeout-less cronjobs).
> >  - Wiki software often includes a manual checkbox to force/prevent
> > notification messages (so either no wait, or infinite wait).
> >  - IM services always operate with zero seconds.
> >  - Funnily enough, I can't remember what the options were for Google
> Wave.
> > I think weekly/daily/hourly?
> >  - Etc.
> >
> > Personally, given Wave's nature, I'm inclined to think this should be a
> > per-wave setting (or per wave #tag, or st). There's no single timeout
> that
> > will satisfy the numerous Wave use cases, so forcing the user to choose
> one
> > (when the bot is added to the wave) miiight be a good idea.
> > Anyway, this is an endemic issue of the Wave concept: so far nobody has
> > come up with a way to differentiate and adapt Wave's behaviour to the
> many
> > different communication platforms it can mimic for each specific wave.
> > Traditional communications forms differentiate themselves by forcing the
> > user to choose different clients each time (chat client vs forum URL vs
> > email software vs social network app vs...). Wave eliminates that barrier
> > but provides no way to build the barrier again when it's needed.
> >
> >
> > Automatically detecting "too big" changes shouldn't be too hard, I
> briefly
> > experimented with it this afternoon: store the plaintext character count
> in
> > each blip's metadata field (the [mailllist-bot?...] string thingie) when
> > the blip is synced; and don't sync again unless the count has changed X
> > percent and/or Y units.
> >
> >
> >
> > As for federation, I have no idea really. I believe that email
> > synchronization is something requested by a big percentage of wave users,
> > so bundling it with wiab by default, and making it easy and
> straightfoward
> > to use, can make a lot of sense for Wave's future. Also, you eliminate
> the
> > dependency from third party servers (I bet most GoogleWave-era bots are
> now
> > offline...).
> >
> >
> >
> > On Fri, Jun 7, 2013 at 11:59 PM, Ali Lown <ali@lown.me.uk> wrote:
> >
> > > Bruno,
> > >
> > > This looks quite cool.
> > >
> > > The main thing I am thinking is how 'big' an event has to be before
> > > triggering sending an email. (A spelling correction is hardly worth
> > > it)
> > > We also don't want a large sequence of emails being sent for changes
> > > happening within a few seconds of each other (think simultaneous
> > > editing of a large wave), so some sort of time threshold will need to
> > > be considered.
> > >
> > > Regarding federation, where should the bot be (presumably on the
> > > server hosting the wave)?
> > >
> > > Anyway, keep up the work on this.
> > >
> > > Ali
> > >
> > > PS. I suspect infrastructure should be able to put in a special rule
> > > to allow this mail if we can designate some 'official' bot from a
> > > particular server.
> > >
> > > On 7 June 2013 22:48, Bruno Gonzalez (aka stenyak) <stenyak@gmail.com>
> > > wrote:
> > > > So I've been working on this for the past days. Still a
> > work-in-progress,
> > > > and will need at least another week of development hours (read: 2-4
> > weeks
> > > > of actual time) before we can really think about migrating to wave.
> > > >
> > > > The apache mailing list is rejecting the emails from my bot, it
> thinks
> > > > they're spam. So for the time being, here's a screenshot-based
> preview:
> > > > http://imgur.com/a/GtGY6
> > > >
> > > > --
> > > > Saludos,
> > > >      Bruno González
> > > >
> > > > _______________________________________________
> > > > Jabber: stenyak AT gmail.com
> > > > http://www.stenyak.com
> > >
> >
> >
> >
> > --
> > Saludos,
> >      Bruno González
> >
> > _______________________________________________
> > Jabber: stenyak AT gmail.com
> > http://www.stenyak.com
> >
>



-- 
Alain Levesque Wavewatchers
Wavyemailbeta:*
*
*Web Page <http://albonobo.com/>
*

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message