You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am in a bind: I have been effectively told by an SNCB that the only 'accepted' method for CRM is to use the online sCRM (https://dmpstats.shinyapps.io/avian_stochcrm/) because all others, stochLAB included, have not been fully tested. But sCRM will not run - within 10-20 seconds of hitting 'run simulation' I am disconnected from the server. I have tried a few times over several days and each time the same result. This is further frustrated by the fact that to try again I have to re-enter most of the inputs each time (despite hitting save).
I have results obtained using stochLAB, run both locally and online which I am confident are robust, but there is a large risk that the SNCB will treat these as unreliable, leading to delays in the consenting process.
I am running 5 species, uploading bootstrapped densities. Any suggestions for how I can resolve this?
The text was updated successfully, but these errors were encountered:
Hi Mark.
Yes - seems there is some problem server-side, we're having a look. It looks like it might be demand on the server (there is a smallish number permitted) - is there some impending deadline that has many folks on at once? It wasn't designed as some heavily used online tool. Bruno has made some increases and we're not getting the disconnect now.
I'll send you an email WRT broader issues of insistence from some areas that older, less reliable software be used, and through a light-use server setup.
I am in a bind: I have been effectively told by an SNCB that the only 'accepted' method for CRM is to use the online sCRM (https://dmpstats.shinyapps.io/avian_stochcrm/) because all others, stochLAB included, have not been fully tested. But sCRM will not run - within 10-20 seconds of hitting 'run simulation' I am disconnected from the server. I have tried a few times over several days and each time the same result. This is further frustrated by the fact that to try again I have to re-enter most of the inputs each time (despite hitting save).
I have results obtained using stochLAB, run both locally and online which I am confident are robust, but there is a large risk that the SNCB will treat these as unreliable, leading to delays in the consenting process.
I am running 5 species, uploading bootstrapped densities. Any suggestions for how I can resolve this?
The text was updated successfully, but these errors were encountered: