I would actually categorize this as a human problem. As a society, we scan documents all the time. Data entry is a massive time sink usually involving lots of corrections.
True code golf would be like switching languages/frameworks for no discernable reason or trying to hit an arbitrary code coverage metric because you're unsatisfied with the current one.
So, like your colleague, I try to strategize delivery. I would never tell a guy not to work on a genius idea that could change the entire paradigm, in fact I let 3 of my guys do just that for the last 3 months, together, sometimes alone. But they either not deliver, or very slowly, or sketchily (not working as well as thought).
So what I do a lot is a compromise. I force them, despite their complaint I don't care enough about the big picture of their non-delivering work, to take smaller issues every week. 1 or 2, give a candy to the user, and do your stuff then.
It worked incredibly well: while they've delivered absolutely nothing of value in their own initiatives yet (but I still think they can, it's just very long term), the users are delighted by a constant stream of little innovations and we've been recently identified as one of the most productive team.
Meanwhile, the team next to us, algo programmers in C++, can take 2 months to enhance an error message. Because they're all revolutionizing the system and have no time for basic human needs. They don't know yet, but they're gone soon. The budget people found there's a new model of delivery that could work, I wonder which one ;)
The industry (and really, it's not specific to the industry) is rife with people who take pearls of wisdom out of their context and apply them without the context to serve their own ends.
Turned her day from hell to cool (she had to filter duplicates by hand and quadratic formulas over 2000+ lines spreadsheet.
That's what computers are for.
Definitely my most cherished code.
Sometimes those damn excel sheets take hours and hours to make even though it would otherwise take me less than a minute to write in SQL or unix. But it pays off every time and pushes a clear message that they can do better.
Any rule of wisdom, followed blindly, leads to disaster.
Don't substitute rules for good judgement.
New volunteers sometimes expect the red-carpet treatment as someone volunteering their (pretty expensive presumably!) time and as someone accomplished in their field. Volunteers usually get the opposite treatment since it turns out the other volunteers tend to come from a similar perspective and someone doing the hard human work is often contributing a lot more to the project than an otherwise excellent engineer.
You can of course choose your volunteer activities to be interesting to you. Open source is one such very fun activity but lo-and-behold in projects I've been involved with the human aspect is often very important and requires a similar amount of hard work. This is often truer for bigger projects/organizations than small ones.
I started coding at 12 because everyone said it was much to hard for me. Now im old and motivate people by doing the same. At the same time its good to be reminded by others what happens if you drop one thing in favor of the other.
As in - perhaps OP and 40+ year person had the same, accurate mental model of the probability distribution of success, but OP and 40+ had different attitudes towards risk/reward.
Or perhaps the OP happened to have a much more realistic picture of the probability of success, and 40+ was unnecessarily pessimistic.
So hard to evaluate these things, and to internalise the fact that other people don't have the information you have all the time.
A nice FOSS library for it was on HN recently, sharing is caring I guess
Went to a presentation at a Golang meetup in Amsterdam by the guys behind this company. Seemed to know their stuff. But I have no real world experience using it.
However, some PDFs are scanned documents, or only contain images. PDFTables doesn't perform Optical Character Recognition (OCR) to turn these images into text.
To process these kinds of documents, you will need to either enable OCR in your scanning software, or run the PDF through specialist OCR software before using PDFTables.
Fast forward a month later, I managed to actually do it, and it worked almost flawlessly, across about 100k scanned PDFs. Multiple long-term projects spawned off from the extracted information, including some projects from the dev who told me to stop.
My point being -- sometimes it seems like that line is used as an excuse to not tackle hard problems if they seem likely to fail. I agree with the line, and it's the exact reason I worked on it as hard as I did, but the dev failed to consider that I understood the social part of the problem, on experienced principle.