Bump to 0.7.3

Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ramalama"
-version = "0.7.2"
+version = "0.7.3"
 dependencies = [
   "argcomplete",
 ]

ramalama/cli.py

@@ -9,7 +9,7 @@ from pathlib import Path
 
 import ramalama.oci
 import ramalama.rag
-from ramalama.common import get_accel, perror, run_cmd
+from ramalama.common import accel_image, get_accel, perror, run_cmd
 from ramalama.config import CONFIG
 from ramalama.model import MODEL_TYPES
 from ramalama.model_factory import ModelFactory
@@ -118,7 +118,7 @@ Needed to access gpu on some systems, but has security implications.""",
     )
     parser.add_argument(
         "--image",
-        default=CONFIG["image"],
+        default=accel_image(CONFIG, None),
         help="OCI container image to run with the specified AI model",
         action=OverrideDefaultAction,
     )

ramalama/common.py

@@ -490,16 +490,26 @@ def minor_release():
     return vers
 
 
+def tagged_image(image):
+    if len(image.split(":")) > 1:
+        return image
+    return f"{image}:{minor_release()}"
+
+
 def accel_image(config, args):
-    if len(args.image.split(":")) > 1:
+    if args and len(args.image.split(":")) > 1:
         return args.image
 
     if hasattr(args, 'image_override'):
-        return f"{args.image}:{minor_release()}"
+        return tagged_image(args.image)
 
-    if args.runtime == "vllm":
-        return "registry.redhat.io/rhelai1/ramalama-vllm"
+    image = os.getenv("RAMALAMA_IMAGE")
+    if image:
+        return tagged_image(image)
 
+    conman = config['engine']
+    images = config['images']
+    set_accel_env_vars()
     env_vars = get_accel_env_vars()
 
     if not env_vars:
@@ -507,8 +517,12 @@ def accel_image(config, args):
     else:
         gpu_type, _ = next(iter(env_vars.items()))
 
-    conman = config['engine']
-    images = config['images']
+    if not args:
+        return tagged_image(images.get(gpu_type, config["image"]))
+
+    if args.runtime == "vllm":
+        return "registry.redhat.io/rhelai1/ramalama-vllm"
+
     image = images.get(gpu_type, args.image)
     if hasattr(args, "rag") and args.rag:
         image += "-rag"

ramalama/rag.py

@@ -71,8 +71,10 @@ COPY {src} /vector.db
             raise e
 
     def generate(self, args):
-        # force accel_image to use -rag version
+        # force accel_image to use -rag version. Drop TAG if it exists
+        # so that accel_image will add -rag to the image specification.
         args.rag = "rag"
+        args.image = args.image.split(":")[0]
         args.image = accel_image(CONFIG, args)
 
         if not args.container:

rpm/python-ramalama.spec

@@ -1,7 +1,7 @@
 %global pypi_name ramalama
 %global forgeurl  https://github.com/containers/%{pypi_name}
 # see ramalama/version.py
-%global version0  0.7.2
+%global version0  0.7.3
 %forgemeta
 
 %global summary   RamaLama is a command line tool for working with AI LLM models

setup.py

@@ -64,7 +64,7 @@ class build_py(build_py_orig):
 
 setuptools.setup(
     name="ramalama",
-    version="0.7.2",
+    version="0.7.3",
     packages=find_packages(),
     cmdclass={"build_py": build_py},
     scripts=["bin/ramalama"],

test/system/060-info.bats

@@ -19,7 +19,7 @@ load helpers
 
     # FIXME Engine  (podman|docker|'')
     tests="
-Image   | "quay.io/ramalama/ramalama"
+Image   | "quay.io/ramalama/ramalama.*"
 Runtime | "llama.cpp"
 Version | "${version}"
 Store   | \\\("${HOME}/.local/share/ramalama"\\\|"/var/lib/ramalama"\\\)