-
Notifications
You must be signed in to change notification settings - Fork 2
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
Add more concise documentation of arguments for the attributes #148
Comments
There a quite a bit of optional arguments. There is already a link to more detailed documentation at http://docs.karrlab.org/obj_tables/master/1.0.8/source/obj_tables.html. |
I realized that the constructor arguments weren't displayed in the documentation. This is now fixed. |
In addition to the API documentation, it would be helpful to have more concise documentation of the arguments to the constructors for the attributes. This could be assembled by parsing the docstrings of the constructors and compiling the docstrings to HTML with sphinx. |
Hi Jonathan, that's a good change, but originally I meant something much simpler. For example, with "Boolean", users may not know what's the required format, eg "True", "true", or "1", or whether several of these are allowed. Similar for lists, the user may not know whether brackets are needed, which separator to use, etc. Of for numbers, whether an exponential notation is allowed, etc. So my idea was to put some examples, for those cases that are not self-evident. But I think it will make the webpage too crowded, so I don't really know where this could be put. Best, |
I see. I could examples to the compiled documentation and then extract that into a separate document. I've tried to make ObjTables flexible. For example:
|
Hi Jonathan,
In the list of datatypes on the website, it could be helpful to give an example string for each datatype (or at least those where strings are not lengthy). In most cases the sytax is obvious, but with Boolean values, lists, or dates, for example, the syntax may be less clear. If this makes the website to crowded (which I think it may do), maybe there could be an extra dataset file that contains one item for each datatype, just for the sake of an example?
Best,
Wolf
The text was updated successfully, but these errors were encountered: