To follow up on my previous posts, I have tried to recreate the OPML check script I created for the Reeder OPML export through LLMs, no luck yet with Claude or Copilot. I know Microsoft Copilot is based on ChatGPT but I was curious if there was a difference in ability, so I logged into the ChatGPT app. Trying the same prompt I had tried with Claude and Copilot. The first result looked very promising, it didn’t reference a R package which was no longer on CRAN, and it managed to process quite a few entries in the OPML file without any issues. It did run into an issue later on but that was easily fixed by a single message to ChatGPT with the error message. Overall I was impressed.
Next step was to reattempt the pipeline wrangling Shiny script. It produced a working version with the initial prompt, I moved on to improvements and through a rather smooth process I ended up with something which was fit for purpose and maintainable.
Having written the pipeline wrangling Shiny script (around 80 lines of R code) the old fashioned way I was curious to see how LLMs would handle it. Having access to MS Copilot and to the free tier of Claude I have tried both of these.
To keep it fair I tried the same prompt with both, and the results were very close for both LLMs, I asked for a partial match for “Won” in a certain field and Claude coded an exact match, both LLMs showed row numbers in the datagrids which I had explicitely asked for not to be shown and in both cases irrelevant old data was shown, due to improper handling of dates.
I just noticed I wanted to start this post with “Well”, which would make 2 in a row. So lets not do that. I did manage to get my Feedreader back under control, more details will follow but it was a fun little excercise in some XML parsing, some automatic checks, followed by some additional manual checks and dealing with a RealmDB (which was actually the first time I encountered this DB type, would have expected SQLite) to get a backup of all the Starred items. It was fun, nothing too complex, I do plan to write it up and try to replicate it using some LLM. Another fun little experiment.
Well it’s been about 2.5 years since my previous entry. Kept thinking I should get back to this but just never seemed to get round to it.
What has happened in the meantime:
I stopped running due to an injury, probably due to trying to hard and not using decent enough shoes, this injury kept playing up when I tried to get back into running again, so after a while I just stopped trying and accepted
About a year after that I joined a gym, rebuild some of my endurance and gained a bit of strength
Took up running again, first on a tread mill and then on the street again. Moved from 5 to 10KM runs
Improved my speed, improved my distances, have run a few runs of more than 22KM
Ran my first 8KM race last Sunday, couldn’t hit the time I wanted, lost a bit too much time in getting past slower runners, but it was fun. Definitely something I would like to do again, which is a good thing because I already signed up for a 15KM race before running the 8KM
So I have gone beyond my expectations, of course the downside is that I now own multiple pairs of running shoes and have clothes I only use for running / the gym. Which also is something I never expected.
On Friday I passed my AWS Solution Architect Professional exam. I did spend some time studying but since technology builds upon existing layers, I could build upon my existing knowledge and experience. Foundational layers related to Business Continuity or DNS always pop up, and having been advising customers about Azure, designing and developing solutions upon Azure for the last 6 years helps with a cloud mindset which is not really that different between AWS and Azure. Of course some things are really different but that’s where incremental progress comes in, figure out where the changes are and study those. No need to start from scratch. Of course this is not just limited to technology, we can often build upon existing layers, skills can be transferable, going from running 0 to 1KM is harder than from 5 to 6KM, a foundation helps.
Today I ran 5KM. And altough it’s nothing special, plenty of people run greater distances each day for me it was a first. My time was actually worse than expected but it didn’t matter, I had proven my past self wrong. For years I was sure I couldn’t run at all, anything above a 100M was simply impossible. But about 7 weeks ago something changed and I just tried running for half a minute. And I went on from there.
I am sure I won’t run really fast and I probably won’t do anything beyond 10KM but it did show me I can be my own worst enemy, when it comes to my mental capabilities I have always had a growth mindset but it turned out it didn’t extend beyond. Thoughts can change. Some limits are hard but don’t let those stop you from trying, there’s usually enough room between where you are and the actual limit.