Merken

DjangoCon US 2015: Lightning Talks

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
what about the end of the time and the and the and the and the and the mn ch alright so my name is Ricardo on in sort of them migrate them like 50 per cent of scientists 50 % of the engineers so I'm going to talk about mutants today so I work for the spallation neutron source of talk all of
this is the picture feature will follow follow this pleasure neutral source for those who don't know the season-opening choke which was all 1 of the Secret cities that was built for the Manhattan project so when you
for that the the lead toward the bond of tall thin over over Japan um worry was was enriched of the knowledge things changed is not secret as anymore in today for example
we have the most powerful computer in the world but we also have huge problems in in in biology in material science we have available science we also have that gets and there's and the parking lot and this is really true because I've seen them again and so just a quick
introduction about spallation neutron source would produce the 2 you will be newtons neutrons they have properties that allow us to study might they have up but they have a 5 it is I always lands that is similar to the distance between atoms so we can see that the matter with Newton's of to produce neutrons with the what we have is to produce hydrogen islands so and not to modify we do electrons we speed these he's beginning isolated roughly 90 per cent of the flight with then chop off the 2 letters because you don't need them and then we would accumulate them in the spring and then we find these GUI of mercury and mercury here of targets and who obtained Newton's like and out those new to the other and directed to instruments I think this is an example of an instrument
that we have here this is repeated against you again see like a little over there are you so but whenever we get those nutrients would have a sample utterance in mock difficult again sample of their so we should know were simple with Newton's those neutrons are scattered and we get them on detectors I'm not going through the delicate Newton's although that so basically only gentle so gentle he's used
everywhere below this process so all the instruments we have I think we have around 12 15 instruments they all have have general somehow also the using of to monitor everything and this is an example of
use these changes to monitor process to use a lot of us great fuzzy plots fuzzy fuzzy stuff right and then there's another thing that we
also do so we do that the reduction because that would make these you'd should have to submit their related to a cluster this is part of the project and I'm working so GABA
something like these we have a reduction in every reduction is composed by several scans we want to minimize the horse so what's what did the users of scientists usually do they measure the same thing several times so for example for a small angle scattering experiment which is basically measuring the deviations of of follower of the scattered neutrons we have the same sample the protein for example can be the solvent in several concentrations so we measure everything we average and we have some data OK so from and then submitted to
his to uh to a to a cluster we can anticipate every election composed of several scans is a symbol MPI job or a single MPI dual or multiple jobs and know this is what I'm working so I
don't have much experience be gentle and I've got a lot of questions on taking advantage of these 5 minutes to get response from you OK so what we have what I have with it is I have my re-election my reflection is composed by several scans and I have the stroll less admit you the reduction or I submit several scans so I have these polymorphic association I have no clue how to do this in general for that uh another thing that I have so I have my results in my results it depends on the inputs of the of the 2 scientists give me and maybe they can be the for these would be Jason but Jason so far is we hold after lunch we don't we only have Jason with the key and the value there's no nested J. some still supported in in in gender so hopes and
summarizing everything so basically I need something like these where I get only objects and I get the pipe associated with each of us can also have something didn't tell so I have my money so the forms that is so that is so before just to submit data the cluttered with a lot of so have like 100 entries in my form or even more sometimes they'll conditional like have the fault sometimes the users think about them and I have like a g query very fast thing that you want the parameters that I have to change in my complaints the act collectively usage and they 1 if I can have these vessel would have to spread the solace package Chris performs better know if it's good or not and I still present
also with some people some like some some um also about about uh Jason ISO balls so the Jason the proper Jason is most supporting general I want this is going to be supporting 1 1 8 5 or only in 1 . no well and that's all I
have for you so please come and help me right thank you very much few so this talk is about sorted containers a module 11 work 9 and this can be stored in containers in 5 minutes my name is Grant tanks and working on pipeline since about 2011 like many of you I was pleasantly surprised and impressed but icons batteries included slowed not
long in the pipeline development asking questions like these had is sort of Python dictionary by had sorted by value and I get the maximum value out of a dictionary could I somehow search keys these questions give you to millions of times for every priority queue were cash for in memory index or even indexable set that somebody needs In response I got some great answers the 1st is ordered as so you might actually also use Django sorted at at 1 point and you quickly learned that ordered doesn't stay sorted it just stays ordered there's also counter which is pretty clever another remain heitinga contraption it's combined with another model called he Q which helps to maintain a binary heap the In all of these were wonderfully
fast as except when they work so they kind of assume that updates can be batched and sorted methods can work lazily and that gets us really far actually like really far but sometimes you want even more it so if C +
+ and Java and . net all have sorted map like datatypes why doesn't thought I found the answer from that Coughlin paraphrasing reader if the user is sophisticated enough to realize that the built-in types find the right solution for the problem then also up to the task of finding an appropriate third-party libraries To that I say good luck but if you
go out there you find a dozen different solutions they all have different API as well have varying levels of compatibility and performance and if we decided only by most downloaded then we'd end up going with list so let's try that the list implements a
B-tree datatype in C with the node size 128 and I was pretty happy with it and so when they had made this discovery and surprise you because the list should shuffle 128 pointers well by shuffle a thousand points but how can bisect the faster so I went back and
I have tried trying try to write something in Python I tried evaluating a lot of different kinds of sorted tree datatypes red black the skip list display there's randomized is treat uh a lot of these are in in the a C + + there's 1 that's a skip list implantation in Python but it was kind of slow all I can say in 5 minutes
is that sorted continues runs with this idea it's kind of like that the tree but it it's only half-heartedly so it relies entirely on the bisect model and while it's slow the program in Python the interpreter the Python interpreter is written in C so if you think of it as programming the interpreter you actually are writing seek and it turns out lists are really fast like really fast trees not so much
so listen to what some smart people had to say about it uh outsmart tellurite good stuff like this simple effective implementation idea of splitting the sorted containers into smaller fragments to avoid that whether and insertion costs just now operates as that last part fastest C extensions was difficult to believe I would need some sort of performance comparison to be convinced this is true the other includes this in the docks it is and Kevin Samuel says and quite a man is not just by the code quality but the actual amount of work you put that is not the documentation benchmarking implantation explanations and it was a true inspiration actually from Jane to be this driven to create that much in terms of stocks lastly and might Summerfield makes a short cliquier rates pipelines batteries included standard library seems to have a battery in scene and the argument that we've never had before his more then it's time that Python offered a full range of collection classes out of the box including sort ones so I submit sort contains thank you if so they all forgetting forgive me for doing create purely promotional talk here and working working yeah um I enthusiast and I might have seen the look over and and I'm really really proud of that you haven't seen it yet come find me and Hello about is an introductory into Django I and II in that non-programmers and designers and people were used to using the sites and so I start out with templates 1st ending static files and check out the website and this is the racial and often we combine off me here and on Amazon as well
for later I want to point out the most helpful critical review which I think is really fun eric says overpriced book ASL publish very introductory certainly not worth 20 dollars at least it's on the jury's paper so what he gets and design major IRA I did everything the companies already and think you miss critical review amid services I'm going really appear on
experiments then again this is why I'm really appear because I wrote this book on and I'm fund-raising for the 2nd but which is hell about Internet introductory concepts so anyone who has gone through a whole and has a basic level or the Austin gender roles tutorial or any other tutorial at launch them into the 1st basic web out and this new book will cover every concepts like adding straight adding an an API working session adding user uploaded images always like things that you would build upon the basic web that and it's going off and I just want to 5 days ago I'm almost at the minimum goal which is fantastic but I am to kick serve to release a Kickstarter's you back exert campaign you'll get the new book by December and if you back the 1st campaign you know a year late from that 1 that will not happen this time I promise that could be in your hands by December the yet how about the concepts this fund-raising on Kickstarter watch the video is to see faces
like this in the video and thank you to everyone who back elastic starter and has been so supportive my journey my book but who few
while gender come you so right my name is dan Deats and up and we give you a quick overview of something that we build called fabric bowl and the title of this talk is look marked no command line and so on it there and wait for the presentation pull up here you title of the false Our 1st all we've got a
company both effect of that supports this project so just give the change point for the company but gender
dashed out 2013 was where we originally developed this piece of software has on the familiar with the competition it has run for a couple years now but we yeah I'd like to say we take we took 1st place but we actually tied for 1st so some what we
try to do is solve some problems that we saw that we have a better tool that's designed help with deployments OK so that there's lots of people that use fabric and I can't get into that but when you're using fabric to manage deployment systems but you need programmers in order to execute that those deployment tests but you also need your programmers have permissions to connect to appropriate machines so they have to have the elevated permissions did you know got your production servers and all I could stop and then you also have to keep track of what's actually happening so when you run this deployment you want widening 1 trapped in know what's being released OK so what we wanted to do
was yell it takes something as safely that you production but without you that so normal workflow could
look like this all kinds of organizations have different workflows but you could be doing programming since and the EQS and that you have then client review but potentially in a staging environment something like that project manager is working with the customs and hey great but is ordered to go we got approved to go on Friday right so it's mandates approved to go at some point in the future so then find rolls around they stop you from working that is your happily working on the next feature and then you run at deployment so that would be great if that project manager or some other non-technical person to go ahead and run that but it's kind when
when we have a workflow that's costly interrupting a developer to run this deployments and manage all of that stuff that have like the driver here having to get out and changes and tires that's that's at least from our perspective
access control on you can have a
system like this we have developers have machines and I have ssh keys in style and then there's SSE he's configured on all of those other machines out there will
be great if we had a system where would have essential place where this restored b has access and we hand up emissions and nice web interface to different of 1st given different levels of access and being able to chat
releases is great so we had a fantastic talk on from 1 to move earlier today where they're talking about human and releases and best practices and uh you know 1 of the unanswered question like I think we release this you know some time last week we wanna get a say that you know 342 on on Monday as 1 1 0 the so what we did was
we built a web interface is a Django project to run your fabric test and so you can create users in the system you can hand up emissions you can manage all that stuff and then you can actually track what happened when this deployments run out so it's
really easy to get started patterns of out but you do your name you migrate your answer and you're up and running and then you
have a beautiful interface that looks like this weapon without all that you will not in and configure your hopes you will configure stages so you'd say OK these Xhosa what I use for doing so you know my testing staging whatever and I've got a stage over here for production and once you actually get
to the point where you wanna deployment would select the stage and then you actually can see over here on the right-hand side there's a droplets to select from the test you'd like to deploy once you go through that
process you click the play you actually see what happen so we have a status that shows that everything was great and use the fallout of everything that happened so all that you've alignment would normally get working on the command line we capture steward cheaper for later so in the event of something bad happened
so we go back in the review OK here's all my deployments they look wonderful but I did have a failure you actually go back and say OK what happened during that failure you know what was the issue but what I really wanted to
do was invited by the help us from the sprints so let me then
taking feedback from people so last nite at the party we want might have some some questions using manually feature like that so will love your input so I can't help as his ideas for features have meetings also are about
the developers and maintain the prizes are here so that those on Friday would image meaning that then know myself that bad unfortunately have been unconscious that not program and then we would do so did want them and then I we've that a Dallas resident the text in a cage added on that I think is the picture is a little bit bigger there the yep everything million here in Texas and
you out on about boat that I know and then of course they get whole project MIT license pull-down test about try it on Standard disorder that OK hello
everyone i'm undergo win again of
I'm his talk not about they they still iterations time but instead about credit I'm working on for the last couple of months as say and so the and this is I was annoyed by the lack of support for things like what's obvious and Jack and I love of circuits that pretty great freedom of
interesting stuff and so people but we having interesting In troubling
that so um so channels is a sort
of idea that I can't we walk back alive solvent 20 now and so what is it is it takes 1 giant is most coal concept changes around the Django use uh but Freeman's talking to request response right everything he doing Django the middle where you're reading the views a reply was responsible for the facts so we try and like about having tasks but the college I do it is still there and so what are the we would that wasn't it 1 of the different abstraction we can take here that still preserves the simplicity agenda but you do stuff with WebSockets with page to be to push the back and hospitals are the stuff that happens kind of sound that old PC 1 request-response and so I took the cold War idea of you takes request returns response and and to good obstruction of sort of set low on his collar consumer lets you know this a channel and text and metadata time and and turns irritable messages now this seems very simplistic and it makes no sense and have a context of Services context and so 1st of all what this means is that you can into interviews consumer because what channels does it make a channel for incoming was the request so you have things about channel it takes the requests dust off them as send that responses on response channels now this is just left plugged into gender your over Jesus but full but if you want to you do that with little stuff but that's the fun part part that as well as all the normal hasty was the stuff you get other things as well so for example channels comes with less visible so you run out out of focus of and then you get lots of channels you can connect to people the condition on so you can say with some of the Mexican state you can just have you talk model when someone with you have custom joules so you can say OK whatever this thing have sites in the message online exp whenever back and channel that's a message you can send stuff that word so and so on and then a football together in the course of the group so 1 of the things you have with what talk in particular is lots of fun and you also the group them but later this people looking at the right level the people looking off from page the child the here and just send messages were more once and so was also group obstruction channels where you go OK we're going to have a better some connects we have a thing missing on the connection of WebSockets IPUs interviews that connects and how's lost and Jim also because you have these custom channels you can do things like OK whenever someone up as you have at all in the same method you put a message on to a different channel and then that tells consumers does farming and a very brief reflection and I'm sure none of you get what's going on here is a cover further and so what will well having landlines splitting Giger pop I taken John which choosy runs a single instance process and make me 3 different layers workflow interface like which is the thing that talks to the acid will see you have your normal was the interface you have lots of interface you have ICT interface in future and other ones and there's a channel like maternal and it's possible routine between the 2 of us side of it and there's a will collide and what happens is that rather than January as a single process that takes a thing doesn't it's time returns the thing the interface layer is sort of handle the negotiated negotiation walls and put those messages onto the channel and then the workers city workers to for example just take 1 1 message time time you run run the appropriate consumer and do the right stuff so the kind of point here is that this is approaching some kind of asynchronous programming right and think programming is great it's really tricky and so my goal is to give the people of Django about to do the useful posit of a sink without having to know things about like a race condition and deadlocks and livelocks and all the other stuff that even with a good framework to a running system really quite off the problems of this stuff so channels is this project and I have the connotations formal values of 2 in our the stage even at my place so um usually get we look entation they chose example and the idea of this project is is meant to be a way to prototypes and could be gender itself in the future so it's probably the Pontiac and install it plug it in and just take over and do stuff is somewhat examples in there as well and and then as the final thing like that of an example project called the of which is kind of like a link aggregator medication thing but would like comments in Chap stuff that so that's also about to to use as a child so we can get by with no if you like and so that that's up seize your works the what's that much of you're much that might be
Quick-Sort
Computeranimation
Besprechung/Interview
Quelle <Physik>
Telekommunikation
Kategorie <Mathematik>
Besprechung/Interview
Speicherabzug
Benutzerführung
Ordinalzahl
Abstand
Computer
Zentraleinheit
Prozess <Physik>
Datenanalyse
Stichprobenumfang
Plot <Graphische Darstellung>
Ordnungsreduktion
Visualisierung
Computeranimation
Web log
Prozess <Physik>
Hash-Algorithmus
Streuung
Mathematisierung
Stapelverarbeitung
Extrempunkt
Computeranimation
Spezialrechner
Stichprobenumfang
MIDI <Musikelektronik>
Ordnungsreduktion
Winkel
Anwendungsspezifischer Prozessor
Winkel
Plot <Graphische Darstellung>
Ordnungsreduktion
Kreisbogen
Konzentrizität
Fuzzy-Logik
Rechter Winkel
Zellularer Automat
Mereologie
Parametersystem
Ablöseblase
Projektive Ebene
Portscanner
Standardabweichung
Resultante
Assoziativgesetz
Spiegelung <Mathematik>
Streuung
Modifikation <Mathematik>
Datenmodell
Einfache Genauigkeit
Winkel
Symboltabelle
Ein-Ausgabe
Ordnungsreduktion
Computeranimation
Spezialrechner
Assoziativgesetz
Multiplikation
Prozess <Informatik>
Geschlecht <Mathematik>
Parametersystem
Endogene Variable
Dualitätstheorie
Ordnungsreduktion
Schlüsselverwaltung
Portscanner
Parametersystem
Mathematische Logik
Güte der Anpassung
Modifikation <Mathematik>
Datenmodell
Multiplikationssatz
Abfrage
Template
Computeranimation
Objekt <Kategorie>
Bildschirmmaske
Bildschirmmaske
Parametersystem
Plot <Graphische Darstellung>
Ordnungsreduktion
Versionsverwaltung
Objekt <Kategorie>
Punkt
Mathematische Logik
Extrempunkt
Regulärer Graph
Textur-Mapping
Extrempunkt
Template
Computeranimation
Informationsmodellierung
Bewegungsunschärfe
Iteration
Endogene Variable
Zählen
Ordnungsreduktion
Softwareentwickler
Normalvektor
Default
Modifikation <Mathematik>
Multiplikationssatz
Datenmodell
Negative Zahl
Instantiierung
Bildschirmsymbol
Quick-Sort
Data Dictionary
Bildschirmmaske
Prioritätswarteschlange
Menge
Automatische Indexierung
Festspeicher
Parametersystem
Speicherverwaltung
Versicherungsmathematiker
Versionsverwaltung
Schlüsselverwaltung
Element <Mathematik>
Applet
Regulärer Graph
Computeranimation
Entscheidungstheorie
Task
Quellcode
Task
Typentheorie
Datentyp
Programmbibliothek
Addition
Implementierung
Modul
Algorithmus
Automatische Indexierung
Binärcode
Vererbungshierarchie
Element <Gruppentheorie>
Paarvergleich
Indexberechnung
Invariante
Mapping <Computergraphik>
Funktion <Mathematik>
Array <Informatik>
Wurzel <Mathematik>
Kategorie <Mathematik>
Programmbibliothek
Speicherverwaltung
Spannweite <Stochastik>
Knotenmenge
Subtraktion
Punkt
Loop
Datentyp
Mailing-Liste
Zeiger <Informatik>
Computeranimation
Übergang
Algorithmus
Partitionsfunktion
Interpretierer
Randbedingung <Mathematik>
Subtraktion
Teilmenge
Datensichtgerät
Default
Schreiben <Datenverarbeitung>
Paarvergleich
Mailing-Liste
Computeranimation
Netzwerktopologie
Quellcode
Informationsmodellierung
Mailing-Liste
Funktion <Mathematik>
Ganze Funktion
Parametersystem
Datentyp
Operations Research
Optimierung
Einfügungsdämpfung
Modul
Web Site
Einfügungsdämpfung
Quader
App <Programm>
Klasse <Mathematik>
Implementierung
Maßerweiterung
Extrempunkt
Term
Code
Computeranimation
Eins
W3C-Standard
Demoszene <Programmierung>
Spannweite <Stochastik>
Programmbibliothek
Weitverkehrsnetz
Flächeninhalt
Maßerweiterung
Modul
Implementierung
Metropolitan area network
Soundverarbeitung
Parametersystem
Konvexe Hülle
Template
Güte der Anpassung
Elektronischer Datenaustausch
Spieltheorie
Paarvergleich
Paarvergleich
Computervirus
Bitrate
Ranking
Quick-Sort
Dienst <Informatik>
Einheit <Mathematik>
Verschlingung
Parametersystem
Heegaard-Zerlegung
Mereologie
Programmbibliothek
Data Mining
Wurm <Informatik>
Inklusion <Mathematik>
Spezialrechner
Benutzerbeteiligung
Geschlecht <Mathematik>
Euler-Diagramm
E-Mail
Extrempunkt
Bildgebendes Verfahren
Videokonferenz
Internetworking
Übergang
Soundverarbeitung
Software
Punkt
Geschlecht <Mathematik>
Mathematisierung
Projektive Ebene
Kombinatorische Gruppentheorie
Softwaretest
Virtuelle Maschine
Weg <Topologie>
Programmiergerät
Datenmanagement
Software
Server
Physikalisches System
Biprodukt
Ordnung <Mathematik>
Hilfesystem
Computeranimation
Programm
Client
Punkt
Datenmanagement
Selbst organisierendes System
sinc-Funktion
Client
Projektive Ebene
Gleitendes Mittel
Programmierumgebung
Computeranimation
Virtuelle Maschine
Softwareentwickler
Druckertreiber
Perspektive
Kanal <Bildverarbeitung>
Mathematisierung
Besprechung/Interview
Gamecontroller
Physikalisches System
Softwareentwickler
Schlüsselverwaltung
Computeranimation
Server
Softwareentwickler
Subtraktion
Benutzeroberfläche
Client
Physikalisches System
Versionsverwaltung
Computeranimation
Übergang
Softwaretest
Schnittstelle
Systemprogrammierung
Objektverfolgung
Task
Benutzeroberfläche
Mustersprache
Projektive Ebene
Physikalisches System
Migration <Informatik>
Computeranimation
Softwaretest
Server
Softwaretest
Punkt
Speicherabzug
Tropfen
Biprodukt
Computeranimation
Homepage
Schnittstelle
Prozess <Physik>
Gewichtete Summe
Ereignishorizont
Computeranimation
Rückkopplung
Rückkopplung
Verbandstheorie
Ein-Ausgabe
Hilfesystem
Computeranimation
Softwaretest
Arithmetisches Mittel
Bit
Projektive Ebene
Entropie
Optimierung
Bildgebendes Verfahren
Computeranimation
Standardabweichung
Digitaltechnik
Computeranimation
Prozess <Physik>
Spiegelung <Mathematik>
Punkt
Atomarität <Informatik>
Formale Grammatik
Gruppenkeim
Gerichteter Graph
Computeranimation
Homepage
Eins
Übergang
Metadaten
Last
Prototyping
Schnittstelle
Verklemmung
Sichtenkonzept
Abstraktionsebene
Güte der Anpassung
Kontextbezogenes System
Dienst <Informatik>
Menge
Rechter Winkel
Geschlecht <Mathematik>
Konditionszahl
Heegaard-Zerlegung
Parallelrechner
Projektive Ebene
Message-Passing
Instantiierung
Aggregatzustand
Subtraktion
Web Site
Ortsoperator
Mathematisierung
Framework <Informatik>
Überlagerung <Mathematik>
Task
Message-Passing
Informationsmodellierung
Endogene Variable
Koroutine
Optimierung
Einfach zusammenhängender Raum
E-Funktion
Einfache Genauigkeit
Physikalisches System
Binder <Informatik>
Sichtenkonzept
Quick-Sort
Mereologie
Wort <Informatik>
Mini-Disc
Avatar <Informatik>

Metadaten

Formale Metadaten

Titel DjangoCon US 2015: Lightning Talks
Serientitel DjangoCon US 2015
Teil 18
Anzahl der Teile 46
Autor Leal, Ricardo Ferraz
Jenks, Grant
Osborn, Tracy
Dietz, Dan
Mitwirkende Confreaks, LLC
Lizenz CC-Namensnennung 3.0 Unported:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/32779
Herausgeber DjangoCon US
Erscheinungsjahr 2015
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract Lightning Talks Ricardo Ferraz Leal "Leveraging Neutron Sciences with Django" Grant Jenks "Python Sorted Containers Module" Tracy Osborn "Hello Web App Kickstarter Campaign" Dan Dietz "Fabric Bolt" Andrew Godwin "Channels Everywhere"

Ähnliche Filme

Loading...