Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Story of the Flash Fill Feature in Excel

blog.sigplan.org

Post reply on HN
Print view View original

21–30 of 49 posts

Re: Story of the Flash Fill Feature in Excel

#21

Post by agumonkey »

CraigJPerry wrote: post #18
agumonkey wrote: post #15

Earlier quoted context omitted.

this is something I'd pay to be addressed in most companies / institutions. Every tool they buy (physical or digital) is always 60% fit, but adding a few bits on top would propel it above 80% and make everybody's days a bliss. Most applications I've seen are designed for generic tasks, except that workers have a very regular set of tasks, if software could be tailored a bit to this (prefill stuff, save clicks) you cu…

This is one advantage of the world switching to webapps over desktop apps. It’s trivial for me to create a chrome extension that enriches or modifies the UI of a bought product. Add an extra menu of common tasks- no problem. Add a datalist dropdown of options to a textbox where the options come from another service dynicamlly? - again no bother but usability of the vendor product went up substantially.

Potentially only, where i work, browsers are managed by IT, no way to patch stuff, no devtools. Also depending on the design, adding your code may never work unless you reverse the whole codebase (the app in place is extremely stateful, all a.href are replaced by procedural js calls).

Don't laugh but I had more results scripting mouse/keyboard automation on 90s oracle GUIs [0] than the app I mentioned above.

[0] I could prefill half the stuff and help the rest with powershell, it gave me 70% time reduction on most tasks

Re: Story of the Flash Fill Feature in Excel

#22

Post by pbhjpbhj »

CraigJPerry wrote: post #18
agumonkey wrote: post #15

Earlier quoted context omitted.

this is something I'd pay to be addressed in most companies / institutions. Every tool they buy (physical or digital) is always 60% fit, but adding a few bits on top would propel it above 80% and make everybody's days a bliss. Most applications I've seen are designed for generic tasks, except that workers have a very regular set of tasks, if software could be tailored a bit to this (prefill stuff, save clicks) you cu…

This is one advantage of the world switching to webapps over desktop apps. It’s trivial for me to create a chrome extension that enriches or modifies the UI of a bought product. Add an extra menu of common tasks- no problem. Add a datalist dropdown of options to a textbox where the options come from another service dynicamlly? - again no bother but usability of the vendor product went up substantially.

I'd be interested in seeing implementation examples of adding 3rd party data from one (3rd party) web source to another for form completion, if you have them please?

Presumably this has to hack around cross-site restrictions nowadays (it's a couple of years since I did any website JavaScript).

Re: Story of the Flash Fill Feature in Excel

#24

Post by pbhjpbhj »

I'm a very occasional Excel user, I can never seem to get it to do what I want on real spreadsheets -- it always trounces my number formats, or background colours, or repeats a short-periodic series instead of creating a series over the whole range. Seems like it could use an idiot mode with a pop-up box with options like "don't change any styles" and "make a series over the whole range" that also gives tips like "use $ in from of a reference to prevent it changing over the series".

Re: Story of the Flash Fill Feature in Excel

#25

Post by Taniwha »

Let me tell you another story about fill in Excel ..... many years ago when the Mac was getting its very first graphics cards I worked for 3rd party hardware manufacturer who had decided to build our first graphics card, in particular 24-bit screen were deathly slow .... people would switch to 8-bit or even 1-bit to use things like Excel and word processors.

I started out intending to design fast vector hardware (because autocad was a thing back then), the end though some serious benchmarking of real world usage meant that we zeroed in on simply making solid fills go really gas (Gbytes/sec, faster than any SGI machine of the time) - one of the big surprises was how fast Excel went with just this hardware speedup - turns out that at times excel would clear a window up to 9 times before ever drawing a useful pixel - all those 'important' black pixels I wanted to make hardware for (character rendering and vectors) were just noise in the traces.

Not just MS: Pagemaker? did something stupid that invalidated the font cache many times a page, Quark? put 2 blank spaces at the end of every line that caused ATM to do stupid N*2 clipping behaviour ..... we could make great hardware and the software guys would just piss it away

Re: Story of the Flash Fill Feature in Excel

#27

Post by TuringTest »

pbhjpbhj wrote: post #24

I'm a very occasional Excel user, I can never seem to get it to do what I want on real spreadsheets -- it always trounces my number formats, or background colours, or repeats a short-periodic series instead of creating a series over the whole range. Seems like it could use an idiot mode with a pop-up box with options like "don't change any styles" and "make a series over the whole range" that also gives tips like "us…

That touches an important point of this kind of "automation magic" in autocompletion tools. They are great when they guess your intent correctly, and terrible when they misfire - specially when, as is common with Microsoft's tools, they are applied automatically and you have to take extra action to reverse them.

It annoys me that they research and apply complex algorithms to analyse the available data, detect the most likely ways to fill the document, build alternative programs to solve the problem and rank which is the most useful; only to then choose one of the inferred options and discard all the rest of the analysis work, getting only a (possibly incorrect) guess of the data completion task.

If, once the work is done, the user were allowed to see what the process was to detect the data, and were allowed to choose between the discarded options or give additional clues as to what the original intention was (the "active-learning session with the user to resolve the ambiguity" that the author of the article discards), the technique could be controlled more precisely and be useful in more situations, not just the ones that work the first time. Giving control to the user would make the technique more robust in my opinion instead of being hit-or-miss.

Re: Story of the Flash Fill Feature in Excel

#28

Post by plorntus »

pbhjpbhj wrote: post #22
CraigJPerry wrote: post #18

Earlier quoted context omitted.

This is one advantage of the world switching to webapps over desktop apps. It’s trivial for me to create a chrome extension that enriches or modifies the UI of a bought product. Add an extra menu of common tasks- no problem. Add a datalist dropdown of options to a textbox where the options come from another service dynicamlly? - again no bother but usability of the vendor product went up substantially.

I'd be interested in seeing implementation examples of adding 3rd party data from one (3rd party) web source to another for form completion, if you have them please? Presumably this has to hack around cross-site restrictions nowadays (it's a couple of years since I did any website JavaScript).

Extensions can request a higher level of permisions so you typically wont run into that issue.

Re: Story of the Flash Fill Feature in Excel

#29

Post by peterburkimsher »

tibbar wrote: post #4

> I went from searching for the hardest problem I can solve to searching for the simplest problem that will have the most impact. This is a fantastic framework to pursue when, for instance, joining a new organization. You can quickly have an outsized impact just by looking for common pain points. As a corollary, it’s very useful to spend lots of time using your own product and talking to the users!

Yes, and it's surprisingly easy to find out where to help! The best colleagues and best teams are where there are a lot of "thank you"s.

The same is true in the article:

"I am indebted in particular to one woman, whose name I will never know". "I also thank the Excel product team for shipping this technology without which this work wouldn’t be as famous."

I wish I knew how to reach out to the author (Sumit Gulwani) or other people in the Excel product team, about a feature that would be really hard to code but really useful to me.

Hierarchical menus of Sheets.

Like bookmarks folders, or Playlist Folders in iTunes, clicking on the Sheet Folder would show every row inside, concatenated into a giant Sheet. Clicking on the Sheet Folder would bring up a menu, which I could then navigate to find a sub-sheet. From a programming perspective I know that it's really hard to do hierarchical nested databases. But from a usability perspective, it would make some of our Excel documents with hundreds of Sheets much easier to navigate.

Re: Story of the Flash Fill Feature in Excel

#30

Post by CraigJPerry »

pbhjpbhj wrote: post #22
CraigJPerry wrote: post #18

Earlier quoted context omitted.

This is one advantage of the world switching to webapps over desktop apps. It’s trivial for me to create a chrome extension that enriches or modifies the UI of a bought product. Add an extra menu of common tasks- no problem. Add a datalist dropdown of options to a textbox where the options come from another service dynicamlly? - again no bother but usability of the vendor product went up substantially.

I'd be interested in seeing implementation examples of adding 3rd party data from one (3rd party) web source to another for form completion, if you have them please? Presumably this has to hack around cross-site restrictions nowadays (it's a couple of years since I did any website JavaScript).

Here's one i made just now: https://gist.github.com/craigjperry2/83b81b7eaf5cf68ca042a9a...

1. Clone that repo

2. Run my fake vendor app - in this case an html page with an empty text box - npx http-server

3. Add the browser extension - tested in firefox - then refresh the page

Now you should see the title fields from https://jsonplaceholder.typicode.com/posts in a dropdown that's now been added to the input field of the fake vendor app page.

Post reply on HN