Extending Functionality of the Symbolic ASR pass#2255
Extending Functionality of the Symbolic ASR pass#2255certik merged 23 commits intolcompilers:mainfrom
Conversation
This reverts commit e478c43.
|
Hello @Thirumalai-Shaktivel Eventually we would like to treat this as the following (basically change the ASR of program 1 to satisfy program 2) Through the latest commit
So now if we apply the pass , we get (just pasting the function body) |
…ing and other things
|
Hello @certik and @Thirumalai-Shaktivel Hence now we can create any composite function from the elementary functions we have and play around with them using assignment and print statements . Let's check As can be seen these match with the commented out assert statements hence I am quite confident of these results. |
|
As of now I haven't refactored any amount of code . As you might see lot of repetetions but that's so that y'all can see all raw code which is being used while reviewing rather than seeing macros etc . I hope that works . |
|
The latest commit implements All results are expected . All asserts pass correctly ! |
|
In the latest commit I've uncommented all symbolic tests and added the EDIT1 : All tests pass through the |
|
@anutosh491 great job!! I think you almost have it. Add the NOFAST and see if things pass. If so, we'll then do a final review and merge it. Yes, it doesn't matter that there is repetition. That can be refactored later. |
|
Ready @certik @Thirumalai-Shaktivel |
|
Great job @anutosh491 ! I can see two avenues how to improve this:
|
|
Thanks for the reviews @certik . Yes now that we have the pass in place , I'll remove the symbolic support from the C backend and then proceed onto refactoring the pass. Another thing I can see is there are quite some issues related to my project that are open and I think now could be closed like |
|
Perfect! Go ahead and assign the label "symbolic", I added it to the above two issues you mentioned. |
No description provided.