Skip to main content
news

Re: spreadsheet ergonomics

owl
SubjectRe: spreadsheet ergonomics
Fromowl
Date04/08/2017 02:24 (04/08/2017 00:24)
Message-ID<ahvz93a.y03@rooftop.invalid>
Client
Newsgroupscomp.os.linux.advocacy
FollowsSnit
FollowupsSnit (50m) > owl

Snit <usenet@gallopinginsanity.com>wrote:

Snit
On 4/4/17, 10:08 PM, in article szvi3.ab@rooftop.invalid, "owl" <owl@rooftop.invalid>wrote:

owl
Snit <usenet@gallopinginsanity.com>wrote:

Snit
On 4/4/17, 9:14 PM, in article ac03ga.hubu32@rooftop.invalid, "owl" <owl@rooftop.invalid>wrote:

owl
Snit <usenet@gallopinginsanity.com>wrote:

Snit
On 4/4/17, 7:05 PM, in article javz903.abu@rooftop.invalid, "owl" <owl@rooftop.invalid>wrote:

owl
How long does it take in Numbers to create a sheet with 52000 tables with locked formulas and labels (A-Z with 2000 tables in each column, each table with a label{#} at the top, and a sum the bottom, summing 10 rows)? And how long does it take to navigate to a specific such table?

2.57 seconds here with sc.

Snit
Already answered this... but, on re-reading I would like to see you do this... and then to show "a sheet" with the tables on it. Would be interesting to see.

owl
I already posted a video of it.

Snit
Missed it. Can you show the link... ONE sheet with tons of tables. Would like to see it.

owl
I smell an argument about tables vs sheets coming on...

Anyway, here's the original vid: https://vid.me/pG0C

Here's a current screenshot with better spacing: http://imgur.com/a/ZhGwv

Snit
Oh, over 24K rows... not just 2000 rows as you say above.

Where did I say "2000 rows"?

And, yes, now I see what you mean by your made-up term "pseudo table". Before I watched the video on my small screen iPod. Also was taking care of a sick child off and on much of the night. Anyway, if you had used, well, even the terms sc uses then that would have made more sense, but I also should have watched the video more carefully.

In any case, yeah, when I push my solution to 24K rows just say it takes a LOT longer than yours. Massively longer. Not just six or even 10 times longer... I can show a video but when I was making it I got bored and started multitasking so I would need to make a new one. :)

Ended up taking it 659 seconds (a second shy of 11 minutes). And even then it did not do the fill-right correctly. I am going to guess I would have to add back a slight pause after the adding of calculations on the first column... so it would talk more than 11 minutes for a working script (well, maybe the pause could be just one second, but whatever... NOT testing tight now).

I think we can safely say even if I got the labels on which reduced the number of SUM functions I added by 1/10th that Numbers would still be so far behind sc in terms of speed on something like this that it should not even be considered. THAT is more what I was expecting... told you I was shocked to get what I thought was about 1/6th your speed.

Rewritten in C. Much faster now. :)

52000 tables, in original format (26 columns, 2000 tables per column, summing 10 rows per table): 0.159 seconds to generate (took 2.57 seconds in bash) less than one second to open

1,404,000 tables in original format (702 columns, 2000 tables): 1.377 seconds to generate (took just over 1 minute in bash) ~6 seconds to open

New format, summing 12 labeled monthly rows filled with random dollar amounts:

52000 tables: 0.644 seconds to generate 3 seconds to open

1,404,000 tables: 964 MB sc file 25 seconds to generate 40 seconds to open 2 seconds to jump from table 0 to table 1403999

http://imgur.com/a/5saYA

Snit (50m) > owl