# The cascade of load() statements and tf_workspace?() calls works around the
# restriction that load() statements need to be at the top of .bzl files.
# E.g. we can not retrieve a new repository with http_archive and then load()
# a macro from that repository in the same file.
load("@//tensorflow:workspace3.bzl", "tf_workspace3")

tf_workspace3()

load("@//tensorflow:workspace2.bzl", "tf_workspace2")

tf_workspace2()

In the runtime package, assembly code can include <code>go_tls.h</code>, which defines
an OS- and architecture-dependent macro <code>get_tls</code> for accessing this register.
The <code>get_tls</code> macro takes one argument, which is the register to load the
<code>g</code> pointer into.
</p>

<p>
For example, the sequence to load <code>g</code> and <code>m</code>

        Map<String, Object> attributes = new HashMap<>();
        attributes.put("source-highlighter", "coderay");
        attributes.put("toc", "macro");
        attributes.put("toclevels", 2);

        // TODO: This breaks if version is changed later
        String versionUrl = DOCS_GRADLE_ORG + project.getVersion();

    visibility = ["//visibility:public"],
)

cc_library(
    name = "c_api_macros_hdrs",
    hdrs = [
        "c_api_macros.h",
    ],
    visibility = ["//visibility:public"],
)

cc_library(
    name = "c_api_macros",
    hdrs = [
        "c_api_macros.h",
        "c_api_macros_internal.h",
    ],
    copts = tf_copts(),
    visibility = ["//visibility:public"],
    deps = [

)

cc_library(
    name = "c_api",
    srcs = ["c_api.cc"],
    hdrs = ["c_api.h"],
    visibility = ["//visibility:public"],
    deps = [
        ":tensor_pjrt_buffer_util",
        "//tensorflow/c:c_api_macros_hdrs",
        "//tensorflow/c:kernels_experimental_hdrs",
        "//tensorflow/c:kernels_hdrs",
        "//tensorflow/c:tf_buffer",
        "//tensorflow/c:tf_status_internal",
        "//tensorflow/c:tf_tensor_internal",

Ce découplage des parties, et le fait d'être un "micro-framework" qui puisse être étendu pour couvrir exactement ce
qui est nécessaire, était une caractéristique clé que je voulais conserver.

Compte tenu de la simplicité de Flask, il semblait bien adapté à la création d'API. La prochaine chose à trouver était un "Django REST Framework" pour Flask.

!!! check "A inspiré **FastAPI** à"

    /*
     * We use double-checked locking. We could the try racy single-check idiom, but that would
     * depend on Logger not contain mutable state.
     *
     * We could use Suppliers.memoizingSupplier here, but I micro-optimized to this implementation
     * to avoid the extra class for the lambda (and maybe more for memoizingSupplier itself) and the
     * indirection.
     *

Complete documentation available at https://docs.deta.sh

Usage:
  deta [flags]
  deta [command]

Available Commands:
  auth        Change auth settings for a deta micro

...
```

</div>

!!! tip "提示"

    安装 CLI 遇到问题时，请参阅 <a href="https://docs.deta.sh/docs/micros/getting_started?ref=fastapi" class="external-link" target="_blank">Deta 官档</a>。

## 使用 CLI 登录

    minisign -Vqm /go/bin/mc -x /go/bin/mc.minisig -P RWTx5Zr1tiHQLwG9keckT0c45M3AGeHD6IvimQHpyRywVWGbP1aVSGav

FROM registry.access.redhat.com/ubi9/ubi-micro:latest

ARG RELEASE

LABEL name="MinIO" \
      vendor="MinIO Inc <******@****.***>" \
      maintainer="MinIO Inc <******@****.***>" \
      version="${RELEASE}" \
      release="${RELEASE}" \

pillow==10.3.0
# For image processing by Material for MkDocs
cairosvg==2.7.0
mkdocstrings[python]==0.24.3
griffe-typingdoc==0.2.2
# For griffe, it formats with black
black==24.3.0