Hacker News new | past | comments | ask | show | jobs | submit login
Little Data: How do we query personal data? (2013) (petekeen.net)
24 points by surprisetalk on March 1, 2024 | hide | past | favorite | 7 comments



I'm a fan on simonw's datasette/dogsheep ecosystem https://datasette.io/


:wave: OP author here! I still haven't found any decent answers to the questions from that post, but I think if I were going to make another concerted effort I would absolutely start with datasette.

Edit: link to the original HN thread with a bunch of interesting comments https://news.ycombinator.com/item?id=6718422


It's interesting to see the founders in that thread mentioning their cracks at it. All seem to be dead now except for HumanAPI (which looks like was acquired by Lexis Nexis).


On personal data, meaning data I control, I have most of my digital life in org-mode, from videos (clickable links running mplayer against the relevant files) to mails (notmuch:/notmuch-search: links) passing from pdfs and mere text notes. For data I do not really control I simply suffer.

My banks offer horrible MANUAL ways to export transactions, two are supported by woob, but it's still a painful crappy situation especially since I'm in EU so they all have OpenBank APIs between them by law (similar to USA OFX feeds) BUT the laws does not mandate opening the access to individual customers and so ALL banks I know in EU choose not to give their customer OpenBank access... There are some companies offer VERY expensive OpenBank access, that's simply too expensive to be an option, so well... I use BeanCount and fava from org-mode notes, but mostly done by hand with templates (Emacs hydras + yasnippet).

Fiscal stuff are even worse, both Italian and French public administrations have horrific and next-to-be-unusable porcals (portals are another thing), automation is somewhat possible, if the Citizen want do read obscene XML specifications that change frequently enough to break simple automation, making such automation no less painful than manual work.

My favorite supermarket offer tickets and invoices from drive via it's portal, unfortunately they are ONLY very crappily formatted pdfs, so parsing them is a painful exercise, with multi-line entries, hard to get tables and so on.

My utilities have pdfs bills, no options for csv/xml or something else. In Italy e-invoicing is mandatory, invoices are in XML, often wrapped in a .p7m envelope to be signed, so I can formally automate them with eventual openssl, xsltproc (apply the xsl style to get and html version, and wkhtmltopdf to get a pdf for human reading and from the xml munging the automated part. Well... Specifications are so horrible that it's simpler (since I do not have many, living in France) to go by hand instead...

My hostings thankfully offer XML and CSV for essentially anything, one also offer a Python library to interact with their invoicing system programmatically.

My IoT setup is an HA mess, too time consuming to manage it properly outside the HA mess.

Long story short IF modern IT were designed for the user we would have good automation and comfort, unfortunately it's not. So we suffer since decades, even if most simple automation was there since the '70s from EDIFACTS and beyond...


love it - another relevant project: https://www.mainframe.so/ - its a bit finicky when i tried, but interesting.


Hello, author of Mainframe here! I appreciate you sharing it!

It's early days for Mainframe. I agree it's still finicky, and I'm working to change that. What made you feel that way about it?

Feel free to reply here or email me. You can reach me through andre [at] mainframe's domain.


Interesting…




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: