2 ms·
While it's cool to see the CS theory used to optimize spreadsheet calculations, it also shows some important points. You can ship the first version with a poor
by phkahler 6y ago
While it's cool to see the CS theory used to optimize spreadsheet calculations, it also shows some important points. You can ship the first version with a poor algorithm that works. Performance improvement matters to users, but after doing what they need correctly. If you don't lead all the way to a solid solution you will be replaced by someone who does.
- Someone 6y agoIk not sure the algorithm used by VisiCalc was poor for the constraints set by its operating environment. Fancier algorithms and their fancy dirty bits and inter-cell links take memory, and that was in short supply (VisiCalc was initially designed to run in 16kB of RAM, but ended up needing 48kB) Also, in most spreadsheets (certainly at the time), evaluating in top to bottom, left to right order is the correct way. That the code had to go through all cells a second time to discover that you’re done didn’t matter, given that there were no other processes that could want to use the CPU, as long as you interrupted calculations on key presses. Also, I’m not sure 123 beat VisiCalc because of its smarter recalculation. I think it were the charts and its macro system (awful by today’s standards, but way better than nothing)