Development of research software at DLR - role and status in practice

Video in TIB AV-Portal: Development of research software at DLR - role and status in practice

Formal Metadata

Title
Development of research software at DLR - role and status in practice
Title of Series
Author
Contributors
License
CC Attribution 3.0 Germany:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Identifiers
Publisher
Release Date
2019
Language
English

Content Metadata

Subject Area
Abstract
Software is an important innovation factor and an integral part of modern research. As a consequence the development of software becomes more and more the norm in research alongside the mere use of software. In an effort to obtain an overview of the current practice of research software development at DLR and to optimize our existing support activities for software developers at DLR, we conducted a DLR-wide survey on research software at the end of 2018. The answers of the approximately 770 participants provide a comprehensive picture of the current state of practice. In this presentation we want to give an insight into the heterogeneous landscape of the research software development of a large distri
Loading...
Software Software Archaeological field survey Simulation Resultant
Service (economics) Transportation theory (mathematics) Variety (linguistics) Archaeological field survey Food energy Field (computer science) Uniform resource locator Estimator Software Different (Kate Ryan album) Software Text editor Spacetime
Ocean current Logical constant INTEGRAL Archaeological field survey Physical law Computer network Bit Expert system Vector potential Number Proof theory Basis <Mathematik> Normed vector space Software Integrated development environment Office suite Electric current
Ljapunov-Exponent Electronic data interchange Software 1 (number) Mereology Mereology Abelian category
Point (geometry) Process (computing) Code 1 (number) Archaeological field survey Code
Performance appraisal Turing test Software developer Computer science Code
Type theory Pairwise comparison Mapping Code Multiplication sign Computer science Right angle Average Computer Number
Shift operator Software developer Software Right angle Shift operator
Residual (numerical analysis) Shift operator Digital rights management Open source Shift operator Digital rights management Resultant
Software Software Shift operator Number
Natural number Average Software Universe (mathematics) Total S.A. Resultant Number
Code Code Right angle
Mixture model Programming language Process (computing) Range (statistics) Code Mereology Position operator
Weight Software Java applet Revision control Control flow Error message Control system
Revision control Axiom of choice Game controller Word Code Revision control Control flow Mereology Number
Revision control Revision control Daylight saving time Control flow Physical system Number
Software Code
Software Code Electronic program guide Software Electronic program guide Code Menu (computing) Software testing Right angle Bit Sinc function
Boss Corporation Software Software testing Student's t-test
Curve Regular graph Software INTEGRAL Decision theory Software Right angle Software testing
Point (geometry) Autonomous System (Internet) Computer program Service (economics) Observational study Link (knot theory) Open source View (database) Decision theory Multiplication sign Mereology Power (physics) Number Revision control Wiki Mathematics Software Position operator Social class Condition number Physical system Pairwise comparison Cellular automaton Planning Bit Digital rights management Arithmetic mean Process (computing) Software Personal digital assistant Website Right angle Arithmetic progression
Point (geometry) Dependent and independent variables Statistics Open source Projective plane Ultraviolet photoelectron spectroscopy Archaeological field survey Number Optical disc drive Digital rights management Word Mathematics Software Hypermedia Musical ensemble Position operator Metropolitan area network
isn't for her and abide aware yeah the roman stages and in
practice which we have for it is the development of research software aren't I will mainly talk about a survey we did so you will perhaps find some similarly questions and answers that we found in the 1st talk today but sometimes you got different results as might be quite interesting anyway z faced for everybody who
doesn't know about Lycée should have because obviously we ask the people and you not see perhaps when I have it's good to have a free said you know what we are how we look like so in total currently we are about editors and 200 people's as the most at as as more than 8 thousand that's correct
we distributed in 20 different locations all over Germany and 40 different institutes and facilities and based on something like a service of lost like in a each institute and facility 1 person to 7 5 like how many how much money do you spend on on developing software and we got I get a debate 20 per cent of the the money which is spent on personal spend suffer development but this is more like an estimation and we also know that to dealers working a lot of different fields from our night takes 2 space t transportation and energy and that there's a lot of different suffer developed developed in all these different fields regarding the 2 different technologies for use team sizes and so on but it gave us a rough understanding what we dealing with that we have a lot of variety and what we did but we didn't know we want to do how would really looks like so we decided we want facts and what we did is we also did a survey
this is based kind of a little bit on the 1 we heard in the 1st today so what we wanted to get as an overview about it's a current role of practice which research suffer developments have with India lot and I especially identify places where we could proof so what we can we office that will help them and it was quite nice we don't 7 hundred 73 tastes just was India law which is quite a nice number and the 1st question which
every we asked everybody's in which Institute do you
work and we saw that
nearly from every Institute regard answers obviously there's some institutes where we get much more but that's also the bigger Institute so it fits quite well
and this was really nice it we got really a good overview about ever saying and the next thing is we
also ask for age and this is something which is really
similar to what we saw in today so nearly everybody's developing the software most of more and the younger years older you get less suffer you develop for me this 2 explanations ones to further your in your career you might work lesson software 2nd and suffer something which is still not so it is a major role of suffers something which is coming up currently so it might be than when you started your career and you know like in the in the later part and suffer was just not a big issue when you start at say perhaps did pick it up just late and
they have then the major we important question is do you write code during a job
plant because having something to do with suffer development 1 saying but writing code to
another I and we got nearly 80 per cent which say gas so even that about 20 per cent who said no but we still wanted to answer the survey 2 runs at this point a so we stopped for those people so we just took the ones
of was an examination who do right Kate I am so we ask them also what's your highest academic qualifications 1 we
saw that this is it's not really perfectly the question but I have from which feel I you you know I we see also that most people of computer science and which is just about
20 per cent so it's much lower than what we saw the morning would whizzing singer was about 40 per cent and then we can also see that nearly as the other high numbers I really reforms engineering facilities which maps really well what we do it DLR so we have mainly engineers so it's not just a computer science right code to people but also as a engineers really writing code and this was really interesting for us if we could see who answered our questionnaire and we have some stuff like politics so it's it's even though some people the the and we wanted to
know OK how does your work we looked like so we ask them in an average month how much time do you spend on something and how much in comparison would you like to spend on something also that type of question already have so
the 1st thing is we also might offer development and we saw that people are spending like 20 to 30 per cent of 40 per cent but they would like to spend more but not a significant shift but there is a shift to the right so this was something I was really surprised about to be honest because are always expected and says the researchers wanted to research and suffer developments just the 2 of them but people who he answered our questionnaire that they wanna do more software development which I think is really really nice
some even sometimes in 1 go to 100 per cent and we also ask how do the 1
how it's about research how much do you do and you wanted a wall and here we see even a stronger shift I'm going we the people wanted
to 45 to 50 per cent and the residue just 20 to 30 before and obviously the stuff they wanted to and that's the same result we got a morning last management obviously people find to do a bit but please not as much as I have to do now and then we have the other activities which sources you wanna do less which is like all the
little things you have to do the in between OK there we
wanted to know how long are you writing software
I so we half their may experience where users can be it's it's a little was trying so like for example for 10 years we have a high number 4 9 and 11 not so much 20 again as a high
number it but at this result in the nature of others is the result of the nature of human saying like how might call underdeveloped Afghan like 10 years so if intent is to know exactly I'm doing 6 years and 3 months in 2 days and the quizzes like What do you kindly kind at university you yeah yeah examines the Yukon this is years or something but even that people was like so your 35 years up to 40 where much of this is real as just like the number but I we have some really Ridgeview's people
there but most people are in the 2 to 10 years said apartment some were really fresh with us and so the average and total of
this is 9 years soft suffer development experience is quite a number and so we ask
like who who uses a code you write the dues for yourself which
is here mostly me it's just so 0 is like the edges right code for most ideal self and 5 is like a just throated for our father people I
and you can see that it's kind of a mixture if you have both extreme to have some people who just fractured for themselves which alpha less people right it just for somebody else so it's it's typical suffer develop position doesn't exist a lot but it exists and but in general people tend to use it for themselves and perhaps some other people they really do it
for their own research in you could say on average and then another and questions interesting question that was in the 2nd part we at the price is on the rise and we saw that T because like 23 per cent of our people develop science and then this C# MATLAB and see and then it cuts that goes always still some job of Fortran as coach is not really a programming language but we can see that we have a wide range of programming languages are used
within DLR and I'm really happy that JavaScript is right quite done but you up ice is really a game-changer error and it's getting more and more which is used by people also question or had hurt
to have already heard today who's using which kind of us Witten control system
and the dead was also multiple choice so we see subversion and his Subversion and git is nearly everything but the frightening part is this number was is 18 comma decimal 3 per cent which it using no kind off version control system and that was
kind of scary for us so we have 112 people of 700 would not even nearly 800 when not using any kind of word in control but they develop code so this is something we need to change but on the
other hand we are happy that like 80 about 80 percent our using subversion get some version is a system we provide generally in daylight innocent away get is used by these people in their own institutes we're introducing it currently so we hope to UPS's number that but we see that we are on the right way here
documentation so now we come a by
what are these people will doing so we also are you determine take active decrementing so you suffer and nearly 90 % that yet and I the 3rd leg right so we asked the what kind of
documentation do today and this is where it gets interesting because the only thing people
do is a right code commons they don't rights developer guides or documentation guides or release notes us a little bit but just not really frequently on inflation notes what you find out is what they do read Meese and this is something where sinc we on the right stop was into using GitLab where we can put in wheat maize so that people would perhaps
Cisco increases there but I long but hexoses documentation for everybody so the same thing we did was testing so again we ask are you testing your software under Gödel and
97 per cent that yes then again I like this can't really read through
I am so we also when do you test yourself and 1st saying and we also like OK dude
development in the red is yes test before you give it to somebody else still mostly yeah that serve some sometimes of printed depends on who you give it to your boss a student like a test for my boss but not for a figurative student a lawyer before I published it's something that should be also like a ready but ever it is like OK we tossed so
again how we do you test your software and then we
are on manually and when we look into a completely automated at on really mind like never so they still clicks
resting so they say the test but yeah it's really just clicking through their own software figuring out if something goes wrong so he is still a lot to be especially automatization we have to help them by providing the right tools to do so where's continous integration but also help them how to do this and then we also looked into references and have decidability off carried and the 1st thing we austere care for making a fair citable and to be honest I was always like curves of people are we enthusiastic about always saying yes sure documenting and testing but for citation I got as its they say also yes but this it may and it's
about 70 per cent is a no don't really care and so we asked but reference the sites software so that 1 was about your own self in it's about bees site as us and that we got at least 60 per cent is a gas study so this is at least 1 thing they they don't care about their and suffer but they do it so again we ask
how do you do it on the here we got like Citybus unjust username 23 per cent username and the link and then around you are licensed of it goes really dark so is this typical of writing my papers name a software and Adafruit night and that's kind of it answers this needs differently to change this is something we want attack and as numbers better and that's it that's what we learned about DLR so there's some changes in comparison to the big ROC questionnaire because I think we have other people or entering but there's some things we see the same and I think we all can learn out of it that this is perhaps what we need to address in our research facilities it might be similar to use and if you have any questions on happy to help you at banks Austria and we do have some time left she made up around 8 minutes to talk about it quickly do we have questioned the yes and did you have a look into whether the people don't use any version control of those 2 have the basic no experience in programming or there's 60 and older interesting question no we have looked into that but we keep these is definitely it would be nice question to know who's in who's not using self as the the written composers but in my experience I just can't be from expense was people I talk to it's also quite some young people who just haven't like started was something they say I'm just writing something for me myself and I just have it about up services is also problems cell so we have a lot of them coming to our classes that are training to offer and dealt with regards to the 1 that said 3 I don't care for their suffered to recital I was wondering how many of those are actually like publishing this often because if you're not publishing yourself and then you of course don't care for it to be citable let that that that that I mean that doesn't mean that people are not in general caring for decidability is just a few not even publishing them now yes that's right but I mean when you publish me like source it or wikis yet and so the problem still is that it's really hard to bulrushes affirming this is like the general problem it gets better currently so we had a good progress but still it's not the same you have people don't care so much about but they will have to change a lot so we know that I'm from talking to people that they want to you but it's just not because we have this condition that you have to read at least 1 paper a year so everything could publish it's always a nice thing to have and even for those people mainly due self-evident system no high to do that yeah for most of my main paper right to researchers more importan then in this case and writing another publication yet so yeah so has to change somebody tries to get in here of sorry question from the management point of view I do not border into the structure of DNA to we'll have the power to make changes to change thinks that you found out that they are not provides and if yes do you have any plans how to start the they're made to reference a bit to the talk to the US Nobel gave and so we are likely the position that we get some Centrify men from central IT so we help introduce armed for example tooling so we really made a number so we we have a lot network where we have access to each institute like 1 person at least and we talked to them like what do you really need be to how can we introduce is that a work for you needs that's what we for example did for GitLab and so was really a whole process was not us deciding when it uses and on the other hand we have the possibility was all of them saying we need this and that we got central IT to bring it into all of the alarm also we introduce we we do training some stuff so we get paid to do this is really nice and we even created some guidelines of stuff we recommend what you do regarding suffer engineering in which tools to use and how to do it and they got part of sufficient quality management handbook which we have where there's an audit every year from every instituted every some years and they get out like have you seen this guidelines how do you deal with his softer you develop does is something we introduced and that really and that and this ordered
dislike management saying and some some of the developers alike to higher ups and they have to answer like who do you follows is if not why not so it's not that you're required to do everything but yes please have to decide where you don't need to to to handle that so in this
point we have a quite good appointee to really start change obviously you have people who want to use it and 1 to change but yet we are in quite happy position there and we currently also so just started a project to have project we also heard about this there was a workshop and singer to talk and wary of people we try to use what we already started over the last years to introduce on a whole hell Maltsev at such a spread out and because when other hamlets and if you say only what you do is nice it also helps us was our people and our management accepting that the something put media sources was long answer I have 1 last question and 1 very quick answer please blue yeah thanks for sharing or this some of which are the red numbers what funding treaty is kind of he got by more than 700 responses and our furor statistics what harmonies are majorities Benham's affordable and I would say like half of them like 3 400 would qualify as research engineers so this is more is is only get but it's more than the entire survey from the odds in the survey that we had before so much which is kind of do you think that these people they see themselves so all s research of a user or or not and why not is do you have any gets man and they I I don't think they see themselves like that almost all bands as some who perhaps if they would so we try to spread the word what is in this but they just see themselves as researchers mainly who also develops a fair and this is the the the problems Yahzee movement is not as began to be honest I hope that more of the Peabody We he was quite a bunch of people giving talks but like attending like being interested in it but we half like workshops and stuff we also could advertise for that so we had a nice position to to advertise people participating and the people there don't feel really to be suffered in a their just see they they perhaps you could say they're like scientific software development said something which would fly better with but for most of them they feel like still being something like a research as something of this so the
Loading...
Feedback
hidden