open TELEMAC-MASCARET The mathematically superior suite of solvers

BK 3.12.7-alpha should no longer contain any header bugs. If it does, please describe what problem you're seeing.

Remember, do not use any of the files created by BK 3.12.6-alpha, as those headers are incorrect. But if you create the same thing with BK 3.12.7-alpha, then the headers should be correct.

Kindly,
Alan

Ever since I've starting taking care of Blue Kenue, I have never saved and then opened an *.in2 file. Good catch, I'll investigate this issue further.

Kindly,
Alan

My development version of BK is now able to load the *.in2 file.



The BK 3.12.8-alpha will have this fix.

Kindly,
Alan

a.barton wrote:
For someone who's on vacation, you're pretty active! Of course, I am hugely grateful for your rapid work, but I hope you get some rest too.
That said, whenever you get a moment to do so, I'd be interested in a CHyMS account. I wish I had enough programming knowledge to help out, but I only know enough to be dangerous in Python

Also, thank you for the updates on the other behaviours (*.in2, *.xyz header, and colour scale). I'll try again soon with the colour scale and see if it's still problematic.

Kind regards,
André

Thanks... although... You're seeing the tip of the iceberg that sinks back 8 years!
A little FYI, in case you're interested. Blue Kenue is written in C++, with a little Fortran sprinkled in one of the libraries. There are currently 12 libraries and Blue Kenue directly links to 5 of them. Green Kenue has some linkage overlap with Blue Kenue, so improvements in one Kenue help out the other (sometimes). One of the libraries has 387 C++ classes; spread over about 906 files (*.h and *.cpp).
Thanks... always appreciate clean, clear bug reports

Any who... back to vacation land! And... good luck with your studies

Kindly,
Alan

Ok... me going a little overboard...

I checked my records and on 30th June 2014, there were 316 C++ classes. So I've added 387-316 = 71 new C++ classes in the last 8 years (till today, 28 October 2022). hhmm... interesting (for me!)

Kindly,
Alan

A word to anyone on the same path as I am (this might warrant its own topic, but someone else may be down this rabbit hole): if you are dealing with SERAFIND format in TELEMAC, you will need to specify the usage of double-precision in your steering file:
Otherwise, you will destroy your stress ball as you wonder what on earth is wrong with TELEMAC, then have a brainwave an hour and a half later as you realize you've seen this somewhere before... It doubles the file size, but it's still better than rounding off to the nearest half-metre!..

Hi,

Just to add a bit of an explanation to the "curious" rounding to half meter accuracy.

4 byte floating point value definition only has provision to store 7 significant digits in the Mantissa.

UTM eastings range from 167,000 meters to 833,000 meters at the equator. Northings range from 0 meters to nearly 9,400,000 in North zones and from just over 1,000,000 meters to 10,000,000 meters in South zones.

So you see, sub-meter precision is simply not possible.
Hence the need for double precision OR you shift the mesh to
an arbirtrary local coordinate system.

Hope this helps... Martin

The next version of the Blue Kenue Reference Manual will mention this topic...



Kindly,
Alan