As part of the Common Workflow Language (CWL) conference this year , @Lexisomes and I'll be conducting a CWL novice tutorial training workshop (APAC region) next week. Register below if you're interested and/or please retweet. Workshop Signup: https://t.co/pmP1hNOZDJ @commonwl

Not too late to sign up for CWLConf 2023! https://t.co/nhbkbtVbMj I'll be in Session 2 (EMEA-APAC) Wed 1 March and Session 3 (APAC-Americas) Friday 3 March. Learn how we're using CWL Expressions at @UMCCR for running workflows on @illumina's Connected Analytics Platform.

. @bitly I want to update the destination URL to ONE of my links. I use this service about one per year. How do I do this without paying $96?

"Writing Advanced CWL JS Expressions in TypeScript" Presenter: Alexis Lucattini @Lexisomes (University of Melbourne Centre for Cancer Research) https://t.co/Y2Qhv1PYMb

New release! @MultiQC version 1.14 just hit the shelves, and it's a big one! 🎉 Massive thank you to everyone who contributed (there were a lot of you!) 🙇🏻‍♂️ https://t.co/U4C9HrYH9V Here's a short thread covering some of the highlights.. 🧵

🎵 It's going down (it's going down) 🎵 I'm yellin' Twitter!

Is it possible to delay the submission of a @GitHub issue until 9am the next business day to make it look like the submitter has a healthy work-life balance? ...asking for a friend I also think this would be a useful feature for reducing notifications outside of business hours.

Why does Outlook send me reminders for meetings that happened a week ago? I clearly attended, I clicked the meeting link through the Outlook calendar at the time of said meeting, and you can't track that? And if I didn't go... no good can come from following up now!

. @cpyne @abcnews Ring lights aren't just for Gen-Z TikTokkers but can help let light through from angles where regular down lights cannot (such as glasses)

.@AusElectoralCom can one do a postal vote without meeting the current eligibility criteria? Given that our current Covid cases are around 50K per day (and rising), that's around 350000 people in isolation (plus close contacts). 'I may be Covid+ in late May' should be an option

10/10 Method 3 cont: This took me days to debug. A few test functions could have picked this up much sooner.

9/10 Method 3: Use tests, for each function you write you should have an expected input / output dataset. It’s no guaruantee it will pick up bugs like this, but you should do all in your power to catch errors before they’re deployed into prod. #Yourprodyourproblem

8/10 Method 2: Use the :: notation to ensure that you’re calling a function from the right package. Rather than loading modules with library that can mask functions from other packages, calling functions from packages means that this is avoided.

7/10 Method 1 cont: If bugs like this are later discovered, the renv ‘lock’ file will be able tell you when that version of the package was present in your code, so long as you version control the lock file.

6/10 Method 1: Version control your packages. renv, (see https://t.co/bBMkyc1wol) is a great way to do so. It also allows for easy upgrade of a single package or all packages and recording the versions of all packages in a 'lock' file.

5/10 You can avoid introducing problematic versions of packages like this into your code in a few ways, see the rest of this thread for some ideas. Also please comment any suggestions below.

4/10 Rapportools hadn’t changed in 8 years until last week, which is when this bug was introduced. So fear not, all of your data is not invalid, likely just a weeks worth.

3/10 I have filed a bug here: https://t.co/046PrW2QIz But the developer has stated he won’t be able to patch the issue anytime soon.

2/10 This is a highly problematic bug as no error is thrown. Having a function throw everything answer off by one is particularly hard to debug. This error essentially has occurred because na.rm=TRUE is not parsed into the subfunction as it keyword argument like it should be.

1/10 So this happened the other day: #R #DataScience