These are chat archives for damianh/Cedar

6th
May 2015
Damian Hickey
@damianh
May 06 2015 07:40
DM, not going to have it as part of example in Projections project
Btw, reverted those name changes.
João Bragança
@thefringeninja
May 06 2015 07:41
k
where are we at with cedar.eventstore
Damian Hickey
@damianh
May 06 2015 07:42
Yves Reynhout
@yreynhout
May 06 2015 08:03
I'll be having a look at Cedar.Domain tonight and dump my feedback here. Cheers.
Damian Hickey
@damianh
May 06 2015 08:23
Cool
Everything is up for discussion.
Damian Hickey
@damianh
May 06 2015 15:41
@thefringeninja Cedar.ES.. soft vs hard deletes, either way, it looks like I have to keep the stream ID around forever?
Yves Reynhout
@yreynhout
May 06 2015 15:41
After scavenging, in GES, it's gone, IIRC. But not entirely sure ...
João Bragança
@thefringeninja
May 06 2015 15:42
yep
Damian Hickey
@damianh
May 06 2015 15:42
Yeah the scavenging means the events are purged, but afait, the streamID needs to remain around to prevent future streams from being created with same ID
Is this correct assesment?
João Bragança
@thefringeninja
May 06 2015 15:43
soft delete just means that I can 'recreate' the stream, except that the EventNumber does not go back to 0
yes
Damian Hickey
@damianh
May 06 2015 15:45
Am going to need a streams table.
Yves Reynhout
@yreynhout
May 06 2015 15:45
Or a marker record ;-)
Damian Hickey
@damianh
May 06 2015 15:46
yar
João Bragança
@thefringeninja
May 06 2015 15:46
hrm. are we going to need soft delete for our use cases?
Damian Hickey
@damianh
May 06 2015 15:47
Prob not, but if the behaviour is you can't recreate a hard deleted stream, I need to replicate that.
Yves Reynhout
@yreynhout
May 06 2015 15:48
why would that even happen? only case I can think of is synchronization.
João Bragança
@thefringeninja
May 06 2015 15:48
damn. a marker record would not prevent you from making two queries anyway
Damian Hickey
@damianh
May 06 2015 15:48
Archiving
yes, sync
Yves Reynhout
@yreynhout
May 06 2015 15:49
no, I mean, why would you claim the same stream id?
Damian Hickey
@damianh
May 06 2015 15:49
if any thoughts or ideas, now is a good idea!
João Bragança
@thefringeninja
May 06 2015 15:49
you shouldn't if you are using a new guid, but that's a fairly big if
Damian Hickey
@damianh
May 06 2015 15:50
anything that generates a stream id deterministically
Yves Reynhout
@yreynhout
May 06 2015 15:51
@damianh feedback around 21:00
Damian Hickey
@damianh
May 06 2015 15:51
i.e. a onfiguration AR
@yreynhout looking forward
and bracing myself lol
Yves Reynhout
@yreynhout
May 06 2015 15:51
why would you archive a config AR?
because of new config?
that would surely mean a new AR or a never ending AR, no?
Damian Hickey
@damianh
May 06 2015 15:52
not sure,
yeah exactly
is there any secenario
so has me wondering why it's in GES too...
I would like Cedar.ES is going to be general , if possible.
Yves Reynhout
@yreynhout
May 06 2015 15:54
meh, I like special purposed better ... but that's just me.
Damian Hickey
@damianh
May 06 2015 15:54
This is going to be a lot less general than NES
Yves Reynhout
@yreynhout
May 06 2015 15:54
Not a big fan of the whole "swappable" part.
I understand
João Bragança
@thefringeninja
May 06 2015 15:55
i think someone asked for soft delete
Damian Hickey
@damianh
May 06 2015 15:55
We have different deployment targets - PCL mobile, onprem sql and ze cloud
João Bragança
@thefringeninja
May 06 2015 15:55
ah, i remember. it's for development
Damian Hickey
@damianh
May 06 2015 15:56
Soft delete is meh for me
So I'll pull that option
João Bragança
@thefringeninja
May 06 2015 15:56
so when you make a mistake and delete a stream you can reuse it
Damian Hickey
@damianh
May 06 2015 15:56
Can't think of any scenarios
João Bragança
@thefringeninja
May 06 2015 15:56
but in production? probably not. if you need to get rid of old events, $maxAge / $maxCount is better
Yves Reynhout
@yreynhout
May 06 2015 15:56
I know, but not sure if that's the same bits and even if, if the infra is the right level of insulation from whatever is beneath.
Damian Hickey
@damianh
May 06 2015 15:56
Hard delete + scavange.
Yves Reynhout
@yreynhout
May 06 2015 15:57
Anyway, gotta run ... TTYIAB (in a bit).
Damian Hickey
@damianh
May 06 2015 15:58
Ok, so hard delete = marks stream for deletion. I can have an Undelete that removes that mark, in the case of mistakes. UNLESS a scavange is run first.
Gives a window
like git gc
Damian Hickey
@damianh
May 06 2015 16:04
and if someone recreates a stream then it'll work in sql but not in GES.
So a wiki - "don't do that".
João Bragança
@thefringeninja
May 06 2015 16:04
i would do anything for love. but not that
Damian Hickey
@damianh
May 06 2015 16:04
what?
Is that an agreement or disagreement?
João Bragança
@thefringeninja
May 06 2015 16:05
it's just a :trollface:
Yves Reynhout
@yreynhout
May 06 2015 18:28
my report will come in a bit later than anticipated (still working on it) ... have to eat first.
Damian Hickey
@damianh
May 06 2015 18:39
no worries
I'm not particularly wedded to it fyi
Regarding the ES. Not going to support soft deletions
Damian Hickey
@damianh
May 06 2015 18:44
But am going to investigate the streamid hangs around forever thing on SQL too.
Don't want that edge case to bite anyone.
Yves Reynhout
@yreynhout
May 06 2015 19:13
A bit harsh, but honesty trumps ass kissing: https://gist.github.com/yreynhout/3d6dacab232c70b4cf98
Damian Hickey
@damianh
May 06 2015 19:14
I'm easily offended.
Yves Reynhout
@yreynhout
May 06 2015 19:14
grow up .... and I'll buy you pack of smokes ;-)
Damian Hickey
@damianh
May 06 2015 19:14
lol
Thanks lad, give me a bit to review it.
Kicking all that out one has to wonder why one would even need a common library.
I tend to agree.
Damian Hickey
@damianh
May 06 2015 19:19
right let me absorb this tomorrow.
Yves Reynhout
@yreynhout
May 06 2015 19:20
no problem ... better to let stuff sink in than to blurt out.
Damian Hickey
@damianh
May 06 2015 19:20
Just FYI, I don't want to couple of the Cedar projects together in any way either.
Yves Reynhout
@yreynhout
May 06 2015 19:20
Why multiple read direction in ES
yeah, I figured that.
seems wise, I take a similar stance on things
the same , not similar
Damian Hickey
@damianh
May 06 2015 19:21
We've identified a couple of scenarios where it's desirable to read a stream backwards. @thefringeninja has some. That is, this isn't a store for just for Aggs to end up in.
Yves Reynhout
@yreynhout
May 06 2015 19:22
which begs the question whether everything should be in the same client api.
Damian Hickey
@damianh
May 06 2015 19:23
This does match GES's client API, as you may have noticed.
The repositories (which I've killed from domain) would define their interfaces, which are then adapted to this?
Yves Reynhout
@yreynhout
May 06 2015 19:24
I did. Yes, seems reasonable.
Damian Hickey
@damianh
May 06 2015 19:24
their dependencies as interfaces I mean
Yves Reynhout
@yreynhout
May 06 2015 19:25
yup. Repo interfaces are owned by the model. Impl. binds ES integration.
Damian Hickey
@damianh
May 06 2015 19:25
yes agree
Yves Reynhout
@yreynhout
May 06 2015 19:25
Even if it's just one interface for the repo.
Damian Hickey
@damianh
May 06 2015 19:25
:+1:
Damian Hickey
@damianh
May 06 2015 19:44
Right beer tiem with the missus.
Yves Reynhout
@yreynhout
May 06 2015 19:58
Santé! Salut! Tchin tchin (not Japanese)!