These are chat archives for exceptionless/Discuss

12th
Apr 2016
Blake Niemyjski
@niemyjski
Apr 12 2016 02:14

@srijken:

// Please reply above this line

From: Feodor Fitsner (Support staff)

Found a bug. Will deploy an update with the fix later today.

Blake Niemyjski
@niemyjski
Apr 12 2016 02:44
I committed a change for regular queues. seems we still get a random failure with hashcodes

Assert.Equal() Failure
Expected: 1
Actual: 2

at Exceptionless.Tests.Plugins.PluginTests.VerifyDeduplicationMultithreaded() in C:\Code\Exceptionless.Net\Source\Tests\Plugins\PluginTests.cs:line 593

Enqueueing event with hash:60276899 to cache.
Adding event with hash:60276899 to cache.
Ignoring duplicate event with hash:60276899
Enqueueing event with hash:519669061 to cache.
Ignoring duplicate event with hash:60276899
Ignoring duplicate event with hash:60276899

I might just merge in your change and take a look
Blake Niemyjski
@niemyjski
Apr 12 2016 02:51
I’m going to bed. I thought it could be the NativeOffset that was causing the difference but it’s not.
would be super nice if we had an AO
AOP way of injecting trace messages around GetHashCode so we could see differences.
wonder if there is something in rosyln
Eric J. Smith
@ejsmith
Apr 12 2016 02:59
They have talked about adding it
Nothing yet
Blake Niemyjski
@niemyjski
Apr 12 2016 03:03
:(
Blake Niemyjski
@niemyjski
Apr 12 2016 03:08
@ejsmith
this is interesting
Screen Shot 2016-04-11 at 10.09.03 PM.png
looks like line number and column are changing
same exact code path (these exceptions are generated in a loop)
    private Exception GetException(string message = "Test") {
        try {
            throw new Exception(message);
        } catch (Exception ex) {
            return ex;
        }
    }
so one offset is at the first { after the argument list and the other is right before the throw
well crap.. can’t really dedup that can we
cause line number and column could be different exceptions
but then again if the method is the same and the type and everything else is the same does line number really matter
thinking we should ignore line number, column number, iloffset and native offset
Blake Niemyjski
@niemyjski
Apr 12 2016 03:15
@srijken easy way to figure out the diff was to do this right above enqueue:
context.Log.Info(typeof(DuplicateCheckerPlugin), context.Resolver.GetJsonSerializer().Serialize(context.Event));
Sander Rijken
@srijken
Apr 12 2016 08:49
@niemyjski and then run a JSONPatch ;)
Blake Niemyjski
@niemyjski
Apr 12 2016 12:18
hehe
yeah
so you think we need that lock?
I think we’d be fine without out
the interlocked I guess I’m fine with keeping that
would cost us just 10x as much as that deal
Sander Rijken
@srijken
Apr 12 2016 13:13
the lock can go
Blake Niemyjski
@niemyjski
Apr 12 2016 13:32
ok
@srijken do you want me to commit that and merge?
Sander Rijken
@srijken
Apr 12 2016 13:34
yeah
Blake Niemyjski
@niemyjski
Apr 12 2016 13:37
done :)
I’ll do testing and get this out there today
thanks :)
Blake Niemyjski
@niemyjski
Apr 12 2016 14:40
https://ci.appveyor.com/project/Exceptionless/exceptionless-net @srijken think I need to reach out to appveyor this was working before.
or put an assembly redirect
only thing I changed was targeting .net 4.5.2
instead of 4.5.1
Sander Rijken
@srijken
Apr 12 2016 14:41
right
what I don't get is what master builds were successful after the MSBuild 14 switch
it had errors that were somehow non-fatal
but the difference between the branch and master was super small
Blake Niemyjski
@niemyjski
Apr 12 2016 14:42
yeah
I know
Sander Rijken
@srijken
Apr 12 2016 14:43
it's the benchmark one that fails
oh wait
Blake Niemyjski
@niemyjski
Apr 12 2016 14:44
yep
Sander Rijken
@srijken
Apr 12 2016 14:44
the benchmark create its own mini solution that gets compiled
maybe it was created on the newer image?
does it clean the working directory every time a build is run?
Blake Niemyjski
@niemyjski
Apr 12 2016 14:44
every build is a fresh vm image
Sander Rijken
@srijken
Apr 12 2016 14:45
oh
ok
so.. then it somehow generates an MSBuild 14 sln/project
Blake Niemyjski
@niemyjski
Apr 12 2016 14:45
PerfDotNet/BenchmarkDotNet#137
Sander Rijken
@srijken
Apr 12 2016 14:46
well.. "If you do not want to wait for the release please Install msbuild 12"
that's what we have :/
Blake Niemyjski
@niemyjski
Apr 12 2016 14:47
Reterget your project to .NET 4.0 (4.0 version of BenchmarkDotNet uses other version of msbuild)
PerfDotNet/BenchmarkDotNet#132
guess I could try moving back down to 4.5.1
not sure how that would chagne anything
should I just skip them for now
sounds like they have some bugs with there stuff
Sander Rijken
@srijken
Apr 12 2016 14:51
maybe that's the best thing
subscribe to the github issues
when they get fixed and a new release is out, try to reenable
maybe create an issue as a reminder to reenable?
Blake Niemyjski
@niemyjski
Apr 12 2016 14:53
yeah
I’m subscribed and I’ll renable it
guess we could create an issue but that’s kind of noise
yay
Eric J. Smith
@ejsmith
Apr 12 2016 14:58
appveyor screwing us, eh?
Blake Niemyjski
@niemyjski
Apr 12 2016 15:00
yeah
Blake Niemyjski
@niemyjski
Apr 12 2016 15:14
I told it to ignore line and column numbmer @ejsmith .. we were getting different results for the same stack trace
Eric J. Smith
@ejsmith
Apr 12 2016 15:15
hmm, weird
Blake Niemyjski
@niemyjski
Apr 12 2016 15:15
seems to be working great now
great work @srijken
I’m shipping it
Eric J. Smith
@ejsmith
Apr 12 2016 15:23
sweeet
thanks @srijken !
I think this will be a massive help when people do things like accidentally deploy broken builds.
won’t eat up their plans in an hour
how do you guys like the release notes
especially the wording and work flow
@BillHenning
Eric J. Smith
@ejsmith
Apr 12 2016 15:58
"with a very high priority"
should be a low priority
Blake Niemyjski
@niemyjski
Apr 12 2016 15:58
it’s 1010
Eric J. Smith
@ejsmith
Apr 12 2016 15:58
high means 1
meaning it runs sooner
Blake Niemyjski
@niemyjski
Apr 12 2016 15:58
sorry
Frank Ebersoll
@frankebersoll
Apr 12 2016 15:59
hello :-)
Blake Niemyjski
@niemyjski
Apr 12 2016 15:59
updated
hello!
Eric J. Smith
@ejsmith
Apr 12 2016 15:59
ola
hola?
Frank Ebersoll
@frankebersoll
Apr 12 2016 15:59
i might have some time tonight and on thursday night
Eric J. Smith
@ejsmith
Apr 12 2016 15:59
lol
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:00
so, if you want to discuss the javascript build, we could do that
Blake Niemyjski
@niemyjski
Apr 12 2016 16:00
sure tonight works :)
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:01
do we have any other recent js issues?
just those
that I’m aware of
need to ensure the latest build of typescript doesn’t introduce any issues because it changed the artifacts a bit
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:05
ok
Blake Niemyjski
@niemyjski
Apr 12 2016 16:11
just let me know what time you are free :)
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:12
in about 3 hours
Blake Niemyjski
@niemyjski
Apr 12 2016 16:13
ok that works for me
Sander Rijken
@srijken
Apr 12 2016 16:17
@niemyjski I'll check later, going home now
next up: adding counts to .js?
Blake Niemyjski
@niemyjski
Apr 12 2016 16:18
yeah
probably would be a good idea
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:18
is the counts thing in regard to event hashes / deduplication?
Blake Niemyjski
@niemyjski
Apr 12 2016 16:18
not sure how much of a pita that would be
yeah
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:18
what we discussed about two months ago?
is this a thing already in the .net client?
Blake Niemyjski
@niemyjski
Apr 12 2016 16:19
yea :worried:
I just shipped it
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:19
why the sad emoji :-)
Blake Niemyjski
@niemyjski
Apr 12 2016 16:19
had to do index changes server side
cause I feel bad that it took us that long to accept a pull request on a feature
guess we needed it to bake and I still need to add the server side stats to make it all light up but it’s all in place now just won’t have the aggregated duplicate counts
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:28
I filed this issue in tsproject: ToddThomson/tsproject#84
Todd is already investigating this
It obviously led to a TypeScript issue: Microsoft/TypeScript#8024
interesting thing is, how almost everything you do with Microsoft related products these days leads to some issue on an official github project somewhere.
still better than connect.microsoft.com :-)
Blake Niemyjski
@niemyjski
Apr 12 2016 16:32
yeah
I hated connect
+1ed it
nice find
so do we need to downgrade?
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:34
it’s just when using minification. which we don’t yet.
Blake Niemyjski
@niemyjski
Apr 12 2016 16:38
well we have the min builds
Frank Ebersoll
@frankebersoll
Apr 12 2016 16:38
But we use uglify.
Tsproject has it builtin since this version
Blake Niemyjski
@niemyjski
Apr 12 2016 16:39
ok
Blake Niemyjski
@niemyjski
Apr 12 2016 17:31
@jmoeltjen can we get that made into a blog post?
Frank Ebersoll
@frankebersoll
Apr 12 2016 19:42
okay!
  1. why gulp?
there was some reasoning against it a week ago or so, but what is the alternative?
Eric J. Smith
@ejsmith
Apr 12 2016 19:44
blake was tlaking about webpack, but that is not applicable to this.
Blake Niemyjski
@niemyjski
Apr 12 2016 19:44
easy to run build task system?
yeah
Eric J. Smith
@ejsmith
Apr 12 2016 19:45
are you having problems with gulp?
Frank Ebersoll
@frankebersoll
Apr 12 2016 19:46
i actually like it.
Eric J. Smith
@ejsmith
Apr 12 2016 19:46
yeah
Blake Niemyjski
@niemyjski
Apr 12 2016 19:46
race conditions of sorts
Eric J. Smith
@ejsmith
Apr 12 2016 19:47
what are you talking about?
Frank Ebersoll
@frankebersoll
Apr 12 2016 19:47
i also looked into webpack and browserify last year, but the results were always much more bloated than what we have with tsproject
Blake Niemyjski
@niemyjski
Apr 12 2016 19:47
were seeing gulp race conditions on the js build and probably our other builds too
Eric J. Smith
@ejsmith
Apr 12 2016 19:48
not using it right. :-)
Sander Rijken
@srijken
Apr 12 2016 19:48
what he says :)
Frank Ebersoll
@frankebersoll
Apr 12 2016 19:48
the race conditions in gulp can be solved using gulp-sequence or so
Eric J. Smith
@ejsmith
Apr 12 2016 19:48
yeah
Frank Ebersoll
@frankebersoll
Apr 12 2016 19:48
or just by having more tasks that depend on each other.
Eric J. Smith
@ejsmith
Apr 12 2016 19:48
var runSequence = require("run-sequence");
yep
Blake Niemyjski
@niemyjski
Apr 12 2016 19:51
eh getting close eric :)
just added user info to all summary data
Eric J. Smith
@ejsmith
Apr 12 2016 19:55
sweet
Blake Niemyjski
@niemyjski
Apr 12 2016 20:01
@ejsmith you know how goofy it’s going to look if the entire event list doesn’t have a user
Eric J. Smith
@ejsmith
Apr 12 2016 20:02
yeah, but it sure will let people know that they should be adding users to events.
:-)
Blake Niemyjski
@niemyjski
Apr 12 2016 20:03
yeah
hehe
Blake Niemyjski
@niemyjski
Apr 12 2016 20:16
eh it’s hidden on mobile
Screen Shot 2016-04-12 at 3.16.17 PM.png
when you hover since it’s truncated it overwrites the existing title tag
it would really take a way from the content area if it’s much bigger
@BillHenning @ejsmith
Screen Shot 2016-04-12 at 3.19.24 PM.png
kinda nice to see user but then also does add a lot of noise
almost be better if there was just a user icon column with a tooltip that had name and identity
thoughts?
Blake Niemyjski
@niemyjski
Apr 12 2016 20:22
@boydpatterson
Blake Niemyjski
@niemyjski
Apr 12 2016 20:27
@frankebersoll @srijken you like that? lol
Sander Rijken
@srijken
Apr 12 2016 20:28
why is it double underlined?
Blake Niemyjski
@niemyjski
Apr 12 2016 20:34
cause the truncate plugin
well good question
what the heck
firefox thing
Blake Niemyjski
@niemyjski
Apr 12 2016 20:39
twbs/bootstrap#16574
Blake Niemyjski
@niemyjski
Apr 12 2016 20:49
known firefox issue will be fixed in the next maintance release
Bill Henning
@BillHenning
Apr 12 2016 20:54
I like seeing the user info upfront and not in an icon tooltip.
What you have in the screeenshot looks good to me
Blake Niemyjski
@niemyjski
Apr 12 2016 20:55
ok
I’m tweaking the sessions view
if you are on all events
and you see a session
it’s only going to show the session id
otherwise you have the user shown multiple times
or do you like that
Bill Henning
@BillHenning
Apr 12 2016 20:56
You mean on "Reports / Sessions"?
Blake Niemyjski
@niemyjski
Apr 12 2016 20:56
Screen Shot 2016-04-12 at 3.56.26 PM.png
no the reports / sessions doesn’t change at all
Bill Henning
@BillHenning
Apr 12 2016 20:57
Ah ok, post what it looks like after the change.
Blake Niemyjski
@niemyjski
Apr 12 2016 20:57
kinda sucks.
anytime it truncates we lose the additonal hover info because the plugin we uses shoves the user display name in the abbr[title] and removes the user id that was in it
well it would just have the 26 digit session id
session: 1233453453453453453453
so maybe it’s good that we leave the full user info in the session sumaries
lol this guy responds asking why don’t I just use bootstrap v4 alpha
LOL
not sure if trolling or serious
Bill Henning
@BillHenning
Apr 12 2016 21:02
Is it still pretty unstable?
Blake Niemyjski
@niemyjski
Apr 12 2016 21:02
niemyjski commented a minute ago
Yes my app is deployed in production and our customers are seeing this with firefox and opera. Also v4 is alpha and will break a lot of stuff on top of it I'm using less.
so are you fine with the way it is
check it out on dev
Blake Niemyjski
@niemyjski
Apr 12 2016 21:09
deploying now
kinda shows off the issues with truncating that column the title that contains the user id is blown away by the truncated display name
Bill Henning
@BillHenning
Apr 12 2016 21:17
Is it supposed to have something longer showing?
In the tooltip I mean.
Eric J. Smith
@ejsmith
Apr 12 2016 21:18
yeah, the tooltip seems like its cut off and also its not showing the user id in the tooltip.
in general though I really like it.
Bill Henning
@BillHenning
Apr 12 2016 21:19
Yeah I like it and am fine taking the tooltip hit to get it.
Eric J. Smith
@ejsmith
Apr 12 2016 21:20
yeah, the display name should be identifying.
and if it’s not then you shouldn’t populate it.
:-)
but the tooltip needs to show the user id as well
since that will be the email address in a lot of cases.
Blake Niemyjski
@niemyjski
Apr 12 2016 21:22
yes
do you read?
:)
"kinda shows off the issues with truncating that column the title that contains the user id is blown away by the truncated display name"
thats trunk8
could move to a tooltip or something
Eric J. Smith
@ejsmith
Apr 12 2016 21:24
figure it out. :-) that’s why you get the big bucks/
Blake Niemyjski
@niemyjski
Apr 12 2016 21:25
apart from that you like everything?
anything that I’m missing?
Eric J. Smith
@ejsmith
Apr 12 2016 21:25
yeah, I like that we are showing it.
Blake Niemyjski
@niemyjski
Apr 12 2016 21:25
ok
you see the stack page?
Eric J. Smith
@ejsmith
Apr 12 2016 21:27
yeah, I really like seeing user on that page.
Blake Niemyjski
@niemyjski
Apr 12 2016 21:27
yeah
Eric J. Smith
@ejsmith
Apr 12 2016 21:27
the user box at the top… don’t love the styling, but it’s ok.
Blake Niemyjski
@niemyjski
Apr 12 2016 21:28
the underline ?
Eric J. Smith
@ejsmith
Apr 12 2016 21:28
yeah
Blake Niemyjski
@niemyjski
Apr 12 2016 21:28
those boxes could use a bit of love at different resolutions
how else would you know you can hover?
Eric J. Smith
@ejsmith
Apr 12 2016 21:29
kind of wonder if we can show all the info without a hover.
Blake Niemyjski
@niemyjski
Apr 12 2016 21:29
it would be a lot
bill has like >100k users
Eric J. Smith
@ejsmith
Apr 12 2016 21:29
yeah, and would probably look out of place.
Blake Niemyjski
@niemyjski
Apr 12 2016 21:29
so showing 104,000 of 105,000 users
Eric J. Smith
@ejsmith
Apr 12 2016 21:29
it’s not horrible… just don’t love the styling.
its good for now.
Blake Niemyjski
@niemyjski
Apr 12 2016 21:30
ok
I got this
:)
Eric J. Smith
@ejsmith
Apr 12 2016 21:30
actually, I just changed the underline color to match the text color and it looks way better.
Screen Shot 2016-04-12 at 4.31.02 PM.png
Blake Niemyjski
@niemyjski
Apr 12 2016 21:31
it’s the same color for me

fff

Eric J. Smith
@ejsmith
Apr 12 2016 21:32
Screen Shot 2016-04-12 at 4.31.50 PM.png
Blake Niemyjski
@niemyjski
Apr 12 2016 21:52
eh
I like white better I think
so tooltips are now: display name (id)
same as sessions format
for hover :)
building now
Eric J. Smith
@ejsmith
Apr 12 2016 21:52
yes, I want it to be white underline
Blake Niemyjski
@niemyjski
Apr 12 2016 21:53
yeah
exceptionless/Exceptionless#211
exceptionless/Exceptionless.UI#86
Blake Niemyjski
@niemyjski
Apr 12 2016 22:04
@jamie94bc did you see they closed that azure issue.. not sure they tried very hard to repo
Blake Niemyjski
@niemyjski
Apr 12 2016 22:18
@ejsmith the only thing I don’t like about putting this content into a abbr title is that it becomes very hard to copy it / select it
would you want to copy an id and search by it?
Blake Niemyjski
@niemyjski
Apr 12 2016 22:27
guess that is a bigger issue to be had
because you can’t search or get access to any truncated content