Progress on loadtxt and savetxt (proposition)#34
Progress on loadtxt and savetxt (proposition)#34jvdp1 wants to merge 0 commit intofortran-lang:masterfrom
Conversation
|
Thanks a lot for this work! My comments:
|
|
The branch is quite messed up. For some reason GitHub shows a diff against the CI file, which it should not. First of all, you should be using a branch in your repository other than master. Here is how you can work. First fix your remotes. Here are mine: So do and if your Then once your remotes are fixed, do from your current master: This will create a branch |
|
@certik Thank you for the procedure. I just followed it, and here is the result: Should I close this PR, and create a new one from loadtxt? |
|
Keep this one open for now. But create a new one from the |
I am afraid that if I do a PR from loadtxt, it will still have all the commits. |
|
If you want, I can push a branch that's polished and you can then pull it and create a PR from it. |
Ok. Let 's do that. |
|
There you go: I used your name and email to create the commit. |
Thanks @certik for your help. Not sure what you did but it worked. See the new PR. |
|
All I've done is to take your branch (master), put it into a new branch |
@certik
*I modified the subroutines for sp to avoid a temporary copy of the input/output array in dp. This approach would have been aslso difficult to extent to qp.
*I added two subroutines for qp.
@ALL:
To be discussed:
*For loading, I implemented two ways: 1)copy-paste of the subroutines and changing the precision and 2)unlimited polymorphic objects. We need to decide which approach to use.
*I would be happy to extent loadtxt and savetxt to integer (1,4,8,16). Is there a way to do it automatically? Or should I repeat it manually (all subroutines will be the same, except for the type of the arrays)?
*Should we add other options, like presence of an header, max columns/rows to load,... (similar to Numpy)?