from Hacker News

Excel Labs, a Microsoft Garage Project

by wlkr on 5/26/23, 9:19 AM with 50 comments

  • by bambax on 5/26/23, 1:52 PM

    Some things in Excel have not improved since 2003, that should be fixed before toying with "AI" . For example, when importing a CSV:

    - Excel cannot guess the encoding of the file, and relies on the user selecting it from a list that has maybe a hundred values (?!); the most common encoding, UTF-8, is neither at the top or at the bottom of that list, but somewhere near the end, and is called "65001 : Unicode (UTF-8)" (the preceding value is "65000 : Unicode (UTF-7)"). There is little chance non-technical users will get this right the first time, or any time thereafter, and the result is files that are circulated with garbled encoding and wrong values.

    - Excel cannot guess the separators either! (How hard can it be?)

    That's probably the reason why one cannot "open" a CSV file directly in Excel and having it displayed properly; one has to go through the whole "import" process. Yet Windows insists all CSVs should automatically open in Excel.

    Yes, it's a minor thing, but it should be so easy to fix; instead of that, recent versions of Office have brought incredibly annoying animations that take 2-3 screens to disable.

  • by heisenzombie on 5/26/23, 10:57 AM

    There’s something unbeatable about the purity of a spreadsheet. For certain kinds of problem, it’s one of the most understandable ways to communicate a program.

    But there are the obvious problems, as formulas get inscrutable and you really want some more powerful data types.

    So I’ve been playing with the new Excel features - lambdas and the new kind of array formulas. And they’re kind of great! I ported some non-trivial analysis algorithms from numpy to excel and it makes for a highly shareable and havkable programming environment for non-coders.

    There’s all sorts of crazy excel warts (I’m doing maths with complex numbers, and the handling of those is a true “WAT”)

    It’s kind of almost-great. I can’t put my finger on it but I feel like it’s close to a really winning programming environment for certain kinds of algorithms-transforming-data programs. I think Excel probably has too much baggage to get there, but these experiments are still really interesting.

  • by lindbergh on 5/26/23, 12:08 PM

    Yes Excel lately is turning into a really compelling product, using functions such as sequence (dynamic arrays), lambda (pure functions) and let (defining sub-variables). You'd think you're coding in a pure functional language, and that's great! Now, ff only Excel had a convolution function! But nonetheless it's a great GUI, everybody knows how to use it and sharing it is dead simple.

    The one downside I can find is the lack of a good plotting library. And yes comments as well is something I miss a lot.

  • by 2Gkashmiri on 5/26/23, 2:13 PM

    I'm one of the libreoffice gang and I rely on =regex as a part of my work on a daily basis.

    Eventually we 'might' get therein terms of new features but for the time being, I am fine.

    Macros are slowly improving in libreoffice but it still ways off anyway.

  • by jbullock35 on 5/26/23, 3:38 PM

    Most of us who use Excel are still using it as though it were 2013. Are there any good tutorials about the major features, like lambdas and dynamic arrays, that have been introduced in the last few years?

  • by rickdeckard on 5/26/23, 11:18 AM

    Oh, I need to check that out. The universal language and portability of a spreadsheet is really hard to beat.

    Seeing how people often read xls-sheets by backtracking formulas, it would be great if MS could add support for comments within formulas (multiline formulas being a thing for many years now)

    That wouldn't blow up the whole complexity of formulas, but still allow at least to explain some of these huge formulas people have to deal with on a daily basis.

  • by Closi on 5/26/23, 10:54 AM

    The advanced formula environment is great, I've been using it for about a year now and it makes LAMDA so much more useful.

  • by null_shift on 5/26/23, 2:57 PM

    I’m assuming the generative AI functions are not running purely locally and would be sharing your data to Microsoft/OpenAI ?