PRPNet 4.0.0 Released

[rogue]

The changes are:

1) Fix a bug in the client where it would delete the genefer checkpoint file, thus causing a test to be started from the beginning after a shutting down the client.
2) Fix a bug in the client where it would report to the server that GFN testing had been completed, even though it wasn't. (This hasn't been tested.)
3) Fixed a bug with work selection that would prevent the server from returning work if it exhausted its cursor of 100 records before finding an appropriate workunit to send.
4) Added onekperclient option. This is used with Sierpinski/Riesel server to ensure that each k goes to one and only one client at a time. This should be used when there are many more k values to be tested than clients.

You can find the new release here.

[MyDogBuster]

I believe that the css needs to be accessible via a link, i.e. http, not as a file.

Link what? What does the line look like in the config file, or is the change in the css file?

I'm just now updating some servers to where the css was introduced.

[rogue]

Originally Posted by MyDogBuster

Link what? What does the line look like in the config file, or is the change in the css file?

I'm just now updating some servers to where the css was introduced.

You should specify an http address for the css, not a file, i.e.,

http://localhost:7100/prpnet.css

instead of

prpnet.css

[Mini-Geek]

Originally Posted by rogue

You should specify an http address for the css, not a file, i.e.,

http://localhost:7100/prpnet.css

instead of

prpnet.css

Also note that PRPnet won't serve up the .css file, so you need to use a port that points to a web server that has the file.

[Mini-Geek]

PRPnet 4.3.5 has a bug in calculating decimal lengths. E.g. 84510*166^19390+1 has 43053 digits, and PRPnet reports 43054 digits. I took a look at the code, in LengthCalculator.cpp line 115: (similar in lines 100 and 105)

return log10(doubleB) * doubleN + log10(doubleK) + 1.0;

The "+ 1.0" should not be there. Without it, you are correctly calculating the log 10 value of the number as a double. When converting the double into an integer, you should take floor(lengthAsDouble)+1 instead of rounding the double to the nearest integer.
I've attached patched files, (both the new files and diff -u generated files) hopefully I fixed it correctly and you can use them.

[rogue]

The changes are:

1) Fix bug where client is unable to extract PRP test time when using genefer.
2) Add links to other web pages for the project to top of every page.
3) For HTML generation, evaluate if GFN divisibility checks can be done and make user_gfns.html available only if true.
4) For HTML generation, make team HTML pages available only if there are any teams are running against the server. This is re-evaluated every 10 minutes.
5) Added estimate for the number of days left of work in the server. This works best if all workunits require similar test times.

You can find the new release here.

[amphoria]

You don't appear to have included a compiled executable for the server.

[rogue]

Originally Posted by amphoria

You don't appear to have included a compiled executable for the server.

Oops. It is now posted with the executable.