rmtfleming
commented
3 months ago Hi supertick,
I wanted to address you by your normal name and I also wanted to understand where you are from. Maybe your gitlab could add a link to your bio?
You are trying to do something that I think is a really good idea. Since a long time, I've wanted to enable those without a matlab licence to use matlab, so by app, I assume you mean this: https://uk.mathworks.com/help/compiler/create-standalone-app-using-application-compiler-app.html
When I took over as lead developer of the cobra toolbox, it was with the ethos that we should not intentionally make an a priori judgement as to what constitutes a "valueable" contribution to the cobra toolbox. We accepted contributions once they followed contribution guidelines https://opencobra.github.io/cobratoolbox/stable/contributing.html Then, it is up to the community to decide for themselves what part of the cobra toolbox to use. As such, there is a variety of levels of quality of code contributions so it will be very difficult to completely isolate all errors.
As part of the developments towards v4 of the toolbox, Farid (in cc) and others have been trying to fix any of the code that fails as a result of running this script https://github.com/opencobra/cobratoolbox/blob/master/test/testAll.m
The idea is to define the core subset of the cobra toolbox by the code that passes the test functions that currently exist. I suggest that you try to isolate that code, then see if that part can be compiled into an app.
One issue that you will face is that the cobra toolbox can manage somewhat with the free glpk solver, but really to make the most of it one needs an industrial solver, like gurobi or mosek, which require licences. How would that work with an app?
By the way, that piece of code function allPaths = DFS(graph, s, e, allPaths = {}, path = [], upperBound = 5) is not in the cobra toolbox: https://github.com/search?q=repo%3Aopencobra%2Fcobratoolbox++allPaths&type=code It might be in one of the submodules, likely this one: https://github.com/dgleich/gaimc/tree/master
It would be an authorship level contribution to v4 if you can get a testAll based app to run, esp with industrial solvers.
Regards,
Ronan
On Wed, 21 Aug 2024 at 17:23, GroG @.***> wrote:
Hello, This is only an observation and question regarding the expectation of master/head to successfully compile with mcc.
I've started trying to compile with with a matlab app. The app code compiles fine, but the cobratoolbox has a variety of errors. Some are invalid default values in functions function allPaths = DFS(graph, s, e, allPaths = {}, path = [], upperBound = 5) But there are a variety of others, still working through them. I'm guessing mcc is not used to validate as part of the release process of cobratoolbox ? Forgive my ignorance, I'm a complete noobie regarding MatLab, however, I do have some software engineering experience. Thanks ! Cheers.
— Reply to this email directly, view it on GitHub https://github.com/opencobra/cobratoolbox/issues/2275, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAQMEOWNRL3DTQD3QKFOK23ZSS5HVAVCNFSM6AAAAABM4MDNCWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGQ3TQNBWGAYDGMY . You are receiving this because you are subscribed to this thread.Message ID: @.***>
supertick
Hello, This is only an observation and question regarding the expectation of master/head to successfully compile with
mcc.I've started trying to compile cobratoolbox with with a matlab app. The app code compiles fine, but the cobratoolbox has a variety of errors. Some are invalid default values in functions:
function allPaths = DFS(graph, s, e, allPaths = {}, path = [], upperBound = 5)But there are a variety of others, I am still working through them. I'm guessingmccis not used to validate as part of the release process of cobratoolbox ? Forgive my ignorance, I'm a complete noobie regarding MatLab, however, I do have some software engineering experience. Thanks ! Cheers.