Hi Rich and Aaron,

Yesterday, I found MSG, which is precisely what I was looking for! I want to test a few grids of synthetic spectra, and I need to interpolate them without losing the flux information (something that I still need to figure out with Starfish, for example, but I think it is impossible).

I just did the Python Walkthrough, and everything worked out fine. I'm downloading the Solar-alpha Göttingen Grids from the website to test them.

I want to import the BTSettl models with their different 'flavors.' I read the documentation and saw the tools to do it. However, I'm new to Python (used to do all in IDL), and I'm trying to figure out how to call the ascii_to_specint function to include the labels, etc, to follow the Importing Data instructions. Could you please help me with that?

The original site of the BTSettl models is not working anymore (I guess after the passing of France Allard the website got removed) but VOSA/VO has them in ascii and VOTable formats, so I plan to use those.

Thanks!
Babs

Following the instructions here: https://msg.readthedocs.io/en/latest/user-guide/quick-start.html#quick-start
the installation fails at the final compilation step. I have tracked it down to some misplaced files in src/forum.

Link to Review thread: openjournals/joss-reviews#4602

...to avoid crashes when symlinks become dangling during execution (e.g., due to filesystem unmounts)

Currently, when loading a specint_t from file, the whole spectrum is read. This is wasteful if the user only wants a subset of the spectrum, and/or is using passbands that cover only part of the spectrum. Allowing for partial reads would alleviate this inefficiency.

Provide a list of python dependencies, with version numbers, ideally in a requirements.txt file. This will make installation much clearer.

Review thread: openjournals/joss-reviews#4602

First, thank you for creating MSG! I'm excited to explore applications of your package.

When comparing the output of PyMSG (following the tutorial in your documentation) with Pysynphot I've noticed what could be a unit misspecification in the PyMSG output. Below is a comparison for a star with Teff = 4803, [Fe/H] = 0.0, and log_g = 4.57 (both spectra are surface flux, not the observed flux at Earth).

Converting the PyMSG output on the right (erg / cm^2 / s / A) to W / m^2 / m requires multiplying by 1e-7 * 1e4 * 1e10 = 1e7, which produces a spectrum 8 orders of magnitude greater than the Pysynphot output on the left. The peak flux on the left matches a black body with the same Teff, so the Pysynphot output looks correct.

I'm wondering if the output units for PyMSG are actually erg / cm^2 / s / cm ? Then the conversion factor would be 1e-7 * 1e4 * 1e2 = 1e-1 and PyMSG would agree with Pysynphot after converting to W / m^2 / m.

Thanks in advance!

...meaning that usual array arithmetic (e.g., multiply by scalar) will not work.

I just tried to build 1.0-rc1 by following the instructions in the Quick Start, specifically:

$ export MESASDK_ROOT=$HOME/mesa/sdk/21.4.1
$ source $MESASDK_ROOT/bin/mesasdk_init.sh
$ tar -xf ~/Downloads/msg-1.0-rc1.tar.gz # after clicking the release link
$ cd msg-1.0-rc1
$ export MSG_DIR=$(pwd)
$ make -C $MSG_DIR

and ran into the error

$ make -C $MSG_DIR
make: Entering directory '/home/wball/code/msg/1.0-rc1'
../src/lib/cmsg_m.fypp:18: error: include file 'forum.inc' not found [FyppFatalError]
../src/lib/fmsg_m.fypp:18: error: include file 'forum.inc' not found [FyppFatalError]
../src/lib/msg_m.fypp:18: error: include file 'forum.inc' not found [FyppFatalError]
../src/lib/forum_m.fypp:18: error: include file 'forum.inc' not found [FyppFatalError]
make[1]: *** No rule to make target 'forum_m.f90', needed by 'forum_m.o'.  Stop.
make: *** [Makefile:15: all] Error 2
make: Leaving directory '/home/wball/code/msg/1.0-rc1'

I noticed that forum is a git submodule in a separate repo. There's an empty folder under $MSG_DIR/src/forum, too.

$ find . -iname *forum*
./src/forum
./src/lib/forum_m.fypp
./build/.forum_m.dpp

The instructions for cloning the git repo and installing the development version do end with a successful build, so I suspect the relevant files from forum aren't being correctly included in the GitHub release archive.

...for consistency with other schemata.

Please give instructions on how to

1. Contribute to the software
2. Report issues or problems with the software
3. Seek support

These can also just be the github features (e.g. github issues)

Review thread: openjournals/joss-reviews#4602

Claims are made for the handling of the graceful handling of large files. I was unable to verify these claims. Instructions on how to set the cache size and what is a reasonable value relative to the system memory would benefit users. Also not apparent what other libraries are supported and how those libraries can be converted to the format required by MSG.

Review thread: #openjournals/joss-reviews#4602

When installing msg, I bug out with the error:

ld: warning: directory not found for option '-L/opt/local/lib'
ld: library not found for -lcfitsio

I have the cfitsio library installed, but I cannot find where to tell the make file to look in the correct directory location. Any ideas?

C documentation could/should be embedded in cmsg.h header file, so Sphinx can pull it out using Hawkmoth or an equivalent extension.

Running

sg = pymsg.SpecGrid('sg-demo.h5')

does work, but running

pg = pymsg.PhotGrid(‘sg-demo.h5’, ‘pb-Kepler-Kepler.K-AB.h5’)

kills the kernel.

I'm running into another bug here. When going through the Python walkthrough, my kernel dies when running:

# Load the SpecGrid
GRID_DIR = os.path.join(MSG_DIR, 'data', 'grids')
specgrid_file_name = os.path.join(GRID_DIR, 'sg-demo.h5')
specgrid = pymsg.SpecGrid(specgrid_file_name)

Specifically, the last line is the culprit. Before I start diving into this, do you have any ideas? Because it's cython, such bugs are quite difficult to find.

Currently there is no automated way to install the software package. This could be fixed by creating a pip compatible setup.py script and uploading the software to pypi.org.

Review Thread: openjournals/joss-reviews#4602

Hi,
I couldn't find any tests. There is nothing in my bin directory for the run_test.sh script to run?

Review Thread: openjournals/joss-reviews#4602

I suggest adding the topic astrophysics in the About section.