[Mlir-commits] [mlir] [mlir][emitc] Refactor the e2e test for TOSA -> EmitC lowering (PR #194686)

[Andrzej Warzyński]

================
@@ -0,0 +1,37 @@
+// DEFINE: %{pipeline} = "builtin.module(\
+// DEFINE:   func.func(\
+// DEFINE:     tosa-to-linalg\
+// DEFINE:   ),\
+// DEFINE:   one-shot-bufferize{\
+// DEFINE:     bufferize-function-boundaries\
+// DEFINE:     function-boundary-type-conversion=identity-layout-map\
+// DEFINE:   },\
+// DEFINE:   buffer-results-to-out-params{\
+// DEFINE:     hoist-static-allocs=true\
+// DEFINE:   },\
+// DEFINE:   func.func(\
+// DEFINE:     convert-linalg-to-loops\
+// DEFINE:   ),\
+// DEFINE:   canonicalize,\
+// DEFINE:   convert-to-emitc\
+// DEFINE: )"
+
+// DEFINE: %{lower_to_emitc} = mlir-opt --pass-pipeline=%{pipeline} %s -o %t
+// DEFINE: %{translate} = mlir-translate -mlir-to-cpp %t -o %t.c 
+// DEFINE: %{compile} =  %host_cc -include stddef.h -c -Wpedantic -Wall -Werror -Wno-unused %t.c
----------------
banach-space wrote:

Thanks for pointing it out!

Using `"` seems like a workaround rather than a fix - how come no other test nor tool requires that? 🤔 The motivating buildbot failure (reported [here](https://github.com/llvm/llvm-project/pull/128439#issuecomment-2679620040)) is also gone, so it's impossible to investigate :(

I am actually tempted to land this without quotes and see what happens - we can always revert. At lest we would gain insight into the buildbot failures. WDYT?

https://github.com/llvm/llvm-project/pull/194686