dfb9a36bb2
pypeline: implement compression
...
Implement compression of objects before they are saved into the storage
provider. Each container type can specify which algorithm to use
(currently `none` or `zstd`) and the compression level.
2026-06-17 10:06:35 +02:00
b1857d18ee
Merge Clift{Function,SingleType}Container
...
The two containers were identical modulo the `Kind`. Use a base class
and make the two subclasses.
2026-06-17 10:06:35 +02:00
3a6ed1e7eb
CliftContainer: accept owning module references
2026-06-15 17:28:22 +02:00
686dc0877a
hasModuleAttr -> isCliftModule
2026-06-15 17:28:22 +02:00
c11cc04768
CliftContainers: standardize context accessors
2026-06-15 17:28:22 +02:00
2da35c1252
Clift: pull all the module creation in one place
2026-06-15 17:28:22 +02:00
bb71b0e5a6
Clift: pull all the context creation in one place
2026-06-15 17:28:22 +02:00
90e5dfd4c9
Pass MLIRContext by address
...
Passing by address is more conventional. This is what MLIR does most of
the time, and it avoids dereferences and addressofs everywhere.
2026-04-30 15:09:16 +03:00
a338194eb7
Move namespace mlir::clift out of namespace mlir
2026-04-27 12:48:24 +03:00
58facb0de2
pipebox: introduce two more clift containers
...
`CliftSingleTypeContainer` for a type-level entity and
`CliftModuleContainer` for a binary-level one (like a header).
2026-03-25 17:26:24 +00:00
b76a09d483
pipebox: rename CliftContainer.h
...
The new name is `CliftContainers.h`.
2026-03-25 17:26:24 +00:00
Giacomo Vercesi