Primenet web design

[Uncwilly]

Originally Posted by James Heinrich

By the time you get up to ~4000M the target TF (according to the current rules, which will undoubtedly be obsolete by then) is 2⁹² (and TF'ing one exponent to that level is ~125THz-days )

GPUto∞orBurst.com

[retina]

Originally Posted by Uncwilly

GPUto∞orBurst.com

GPUto∞orBurstIntoFlames.com more likely. Those things run HOT.

[Madpoo]

Originally Posted by potonono

Notice: Undefined variable: etacolor in C:\inetpub\www\assignments\default.php on line 126
Notice: Undefined variable: etacolor in C:\inetpub\www\assignments\default.php on line 130

Yeah, I fixed that on my work-in-progress version of the page. I forgot that sometimes the ETA is blank and didn't have a default color set.

My work-in-progress version of the assignments page now has a tooltip pop up showing the computer name when you hover over the account name for an assignment. That seems much better, to me anyway. I considered doing the same for some of the dates but just removing that computer column really seemed to help things out so I might call it good, or good enough.

FYI, it's kind of weird, seeing people who haven't even started work on an assignment for 300+ days and yet their computer is checking in every few days with an updated completion date. But I guess that's a topic for another thread. Just seems like if they haven't even started on it yet, merely checking in saying "yeah yeah, I'll get to it" should result in some kind of reassignment according to the new rules but maybe those were grandfathered in.

[Uncwilly]

Originally Posted by Uncwilly

The problem is here:

<th>How far factored<br>(2^n)</th>

That <br> causes the (2^n) to become its own line.

I see that that has been addressed in the default version. I had to reset my webquery to the new table placement/name. Then I had to chase down all of the changes to my macros that I had been working on to adapt to something else. If I was doing it all new I might take the XML option. Everything is stable now on this end.

BTW, I noticed that the new version of the page is a bit slower than the initial version (after changing to the new server), but nothing to complain about. My macro run time is now about 14 seconds vs. about 8 before.

[cuBerBruce]

Originally Posted by Madpoo

My work-in-progress version of the assignments page now has a tooltip pop up showing the computer name when you hover over the account name for an assignment. That seems much better, to me anyway. I considered doing the same for some of the dates but just removing that computer column really seemed to help things out so I might call it good, or good enough.

The thing I dislike about moving information to tooltips is that when I copy/paste a table into Excel, the tooltip information is lost. Because we have a new, faster server, must we have reduced functionality than what we had before?

It just seems to me that this effort to reduce functionality in order to "force fit" tables into a somewhat arbitrarily chosen fixed width of space is a step in the wrong direction. On my MacBook, about 40% of the browser's available width is simply wasted, while the Recent Results page (admittedly something I almost never look at) requires a horizontal scrollbar despite the fact that there is plenty of real estate in the browser window to show the whole width.

I hope we can at least have some sort of option of getting the "full" information we could get before, rather than being forced to get less.

Some time ago, date/time information in the Account Assignments Details page was reduced to merely showing date information. As a result, sorting by the "Estimated Completion" column is often times not sorted "correctly." This is obvious when the "days to go" column is out of order after sorting by the "Estimated Completion" column. Alternately, sorting by "days to go" will often result in the "Estimated Completion" being out of order. When the completion date had time of day information, sorting by that column would pretty much always give you the correct order (assuming the predicted times to be accurate, which is another thing, of course).

[Madpoo]

Originally Posted by swl551

Please consider...

...

What would really be cool is a version of the stats page that returned XML so I don't have to screen scrape then convert to XML on my side.

  <StatsTable>
    <id>0</id>
    <rank>1</rank>
    <person>THEJUDGER</person>
   <credit>1204413.201</credit>
  </StatsTable>
  <StatsTable>
    <id>1</id>
    <rank>2</rank>
    <person>NEVER ODD OR EVEN</person>
    <credit>906154.351</credit>
  </StatsTable>

I think something like that is possible. Once I've ironed out a few kinks in some other pages I really would like to add XML options to various things. Then when people say their scraper broke because of changes on page XYZ I'll redirect them to the nice XML option, designed for our new robotic overlords (ahem, I mean just plain bots), so that we human slaves (ahem, I mean humans) can enjoy the pretty version.

[Madpoo]

Originally Posted by Uncwilly

I see that that has been addressed in the default version. I had to reset my webquery to the new table placement/name. Then I had to chase down all of the changes to my macros that I had been working on to adapt to something else. If I was doing it all new I might take the XML option. Everything is stable now on this end.

BTW, I noticed that the new version of the page is a bit slower than the initial version (after changing to the new server), but nothing to complain about. My macro run time is now about 14 seconds vs. about 8 before.

The table markup looks nice to humans but adds page weight if you're trying to pull data out.

I think you would be better served with a plain-text (CSV) or XML format. That's another one to add to the "lets XML that thing" pile.

I'm kind of surprised, but maybe I shouldn't be, at how much scraping of reports was going on. I mean, that's great, because there's some good data gathering and presenting going on, it's just not something I really even thought about.

[Madpoo]

Originally Posted by cuBerBruce

The thing I dislike about moving information to tooltips is that when I copy/paste a table into Excel, the tooltip information is lost. Because we have a new, faster server, must we have reduced functionality than what we had before?

The tip info is there in the HTML markup at least, as the "title=" for that <td> cell. I don't know if Excel specifically can do anything with that since the cut/paste from a table is probably only doing <th> <td> contents.

Originally Posted by cuBerBruce

It just seems to me that this effort to reduce functionality in order to "force fit" tables into a somewhat arbitrarily chosen fixed width of space is a step in the wrong direction. On my MacBook, about 40% of the browser's available width is simply wasted, while the Recent Results page (admittedly something I almost never look at) requires a horizontal scrollbar despite the fact that there is plenty of real estate in the browser window to show the whole width.

Yeah, the trouble is, people in general don't have a good experience when reading a paragraph in widescreen. That's why books are portrait mode. :) There's some usability studies out there (none of which I have ever read of course) that are the reason most websites will restrain the content to a particular section, because that's how we read.

This forum software itself is a case in point... sure, when responding right now I have a narrow 527 pixel window to type my reply, which is great. It's readable (and almost too narrow, I think). But when reading messages, everything is 1600 pixels across or whatever (I'm RDP'd into a machine at 1920, so I'm sure it's wider if I'm viewing it natively). We shouldn't enjoy scanning back and forth that much when going from one line to the next.

Originally Posted by cuBerBruce

I hope we can at least have some sort of option of getting the "full" information we could get before, rather than being forced to get less.

Some time ago, date/time information in the Account Assignments Details page was reduced to merely showing date information. As a result, sorting by the "Estimated Completion" column is often times not sorted "correctly." This is obvious when the "days to go" column is out of order after sorting by the "Estimated Completion" column. Alternately, sorting by "days to go" will often result in the "Estimated Completion" being out of order. When the completion date had time of day information, sorting by that column would pretty much always give you the correct order (assuming the predicted times to be accurate, which is another thing, of course).

If it's useful info, I'll leave it as is. In the grand scheme of things I wasn't sure how useful the computer name (or "CPU name" in the software's parlance) would be to anyone, except the account holder. From a privacy perspective I'm not entirely comfortable with sharing those names publicly anyway. Not like computer names are trade secrets or whatever, but a large account or team with lots of computers, the naming scheme might be useful to ne'er do wells, but that could just be me.

The table sorting plugin I'm using (good old jquery tablesorter.js) can sort based on some other data besides the cell content. I'm using that to sort by the % complete on the "new" assignments page, even though the visible data says "LL, xx%" or just "xx%" depending. I added just the % itself to an attribute and told tablesorter to use that, and it works great.

Something similar could be done with date/time for sorting purposes... just show the date so it's not cluttering up my eyeballs but the time is there to sort on.

It's like I've been saying though, pretty human tables make lousy robot readable data. As long as the data is being parsed by the server to generate fancy pants views, it's not too much harder to present it as XML as well so that people grabbing these stats won't have to work so hard at it, plus the XML versions will be totally minus all of the other HTML stuff (header, footer, javascript, etc) and would present less load on the server... which is probably good since bots will crawl way more pages than humans tend to.

If you're looking for "the firehose" of data, don't expect that in an HTML table since that'll be ugly as anything... if you want that much data, you're probably doing something with it and I'd just as soon generate the raw data so you can import to Excel or whatever and tweak it for your own personal needs.

[chalsall]

Originally Posted by retina

GPUto∞orBurstIntoFlames.com more likely. Those things run HOT!

Indeed they do.

To answer the direct question, it's going to be gpu72.com for a while.

[Madpoo]

Originally Posted by chalsall

Indeed they do.

To answer the direct question, it's going to be gpu72.com for a while.

Aha! But don't forget quantum computers are coming... like, soon or something. :) They'll find the next 1000 Mersenne primes while booting up.

[TheMawn]

Originally Posted by Madpoo

Aha! But don't forget quantum computers are coming... like, soon or something. :) They'll find the next 1000 Mersenne primes while booting up.

No. They will simultaneously find yet not find the next 1000 Mersenne primes while simultaneously booting up and not booting up.