should the constructor for the main contract appear in the dispatch table? #427
Comments
ivoysey
added a commit
that referenced
this issue
Feb 25, 2022
ivoysey
added a commit
that referenced
this issue
Mar 29, 2022
) * first pass at including default constructors * adding a time string to the benchmarks file so it doesn't get overwritten * updating the test script to optionally save the optimized and pretty yul from running solc, for local debugging * generate constructor and arguments * changing the name of the test transaction to avoid #424 * adding a prefix to constructor argument names so that they are unique * whitespace, scala style * writing a helper for unified names of transactions * updating codegen to use the single helper, adding comments * convenience script for local debugging * updating translateTransaction to be able to use the unified helper for transaction names * fixing whitespace * changing the test script so that it calls transactions from main with the contract name prefix * removing inMain boolean argument, it was a symptom of nonuniform transaction names. this addresses #407 * removing some redunant code in translateProgram, pushing helpers into util file * fixing a silly error, changing how line comments get printed for readability * first pass at producing the sequence of assignments for the body of the default constructor * whitespace, scala style * some comments, changing the signature of the internal function so i can actually use fold * removing an accidental implementation of seq.reverse * updating helper script * update the field where it actually is in memory; adding a helper function because this code alsp appears in the assignment case * using the helper to declare vars not just assign them * adding a test specifically for default constructors * whitespace * another missing reverse here * adding comments * first cut at calling the constructor for main per #426 #427 * comment * adding memory allocation to the invoke block, calling default constructor and tracer there. first cut * moving the codecopy after the tracer def and call * adding a test case * scraps * chainging boolean literals to be integers rather than the typed constants, per christians sugguestion * work on storing pointers and following them in constructors * rewrite pointers as they get copied * whitespace, style * adding a shorter constant * adding a commandline flag for benchmarks, and also a check for dumping them to std out in CI * env vars * benchmarks in CI * benchmarks in CI * adding more general log comparison code * logging updates * scraps: changing the notion of this for the dispatch table to be the first address in storage, removing call to the tracer after new * fussing with test definitions * checking for storage vs memory on field loads as well as stores. comparing (x > threshold-1) so that we look in the right place for the first field in the first slot in storage * subtracting from the threshold not the address * without calling the trace after the constructor, we don't expect invoked logs * this fixes a few tests * offset into addresses in storage * whitespace, scala style * clean up * adding a test case * removing excessive gas from two tests, updating a comment per review * trailing commas
ivoysey
added a commit
that referenced
this issue
Apr 1, 2022
* first pass at including default constructors * adding a time string to the benchmarks file so it doesn't get overwritten * updating the test script to optionally save the optimized and pretty yul from running solc, for local debugging * generate constructor and arguments * changing the name of the test transaction to avoid #424 * adding a prefix to constructor argument names so that they are unique * whitespace, scala style * writing a helper for unified names of transactions * updating codegen to use the single helper, adding comments * convenience script for local debugging * updating translateTransaction to be able to use the unified helper for transaction names * fixing whitespace * changing the test script so that it calls transactions from main with the contract name prefix * removing inMain boolean argument, it was a symptom of nonuniform transaction names. this addresses #407 * removing some redunant code in translateProgram, pushing helpers into util file * fixing a silly error, changing how line comments get printed for readability * first pass at producing the sequence of assignments for the body of the default constructor * whitespace, scala style * some comments, changing the signature of the internal function so i can actually use fold * removing an accidental implementation of seq.reverse * updating helper script * update the field where it actually is in memory; adding a helper function because this code alsp appears in the assignment case * using the helper to declare vars not just assign them * adding a test specifically for default constructors * whitespace * another missing reverse here * adding comments * first cut at calling the constructor for main per #426 #427 * comment * adding memory allocation to the invoke block, calling default constructor and tracer there. first cut * moving the codecopy after the tracer def and call * adding a test case * scraps * first cut at tracing after pointer write * adding a couple new helper functions; changing the update field helper to add the storage offset, which I think breaks the SetGetDefaultChecks test. * using new helper functions. adding tracer call code in pointer writes * reversing a bad change, adding a comment about why so i dont feel tempted to do it again * after debugging, changing which identifier goes to the trace call * whitespace, comments * Removing trace this test case. see #434
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
more generally, look at which transactions appear in dispatch and do a little house cleaning. see also #370, #385.
i think i can send over a sequence of transactions paired with a boolean for if they ought to appear in the table or not, from codegen to the yul ast writer.
The text was updated successfully, but these errors were encountered: