跳到正文
历史记录大型仓库

为极速加载大型仓库 而生

gitā 使用 Swift 和 AppKit 编写,每次只向 git 请求即将显示的内容。Linux 内核拥有约 150 万次提交,但在 gitā 中打开和滚动就跟普通仓库一样轻快。

立即体验

AgentNone
BranchmasterUp to date
Pull requestNone
ChecksNone
Preview deploymentNot linked
HistoryAuthorDateSHAStatus
drm: widen the counter to 64 bitsPriya Nair3m ago071c431
nvme: widen the counter to 64 bitsJonas Lindqvist12m ago09d3248
Merge pull request #9800 from locking/next-0Jonas Lindqvist21m ago8e7b915Checks passed
tcp: widen the counter to 64 bitsMira Okafor30m ago890dce7
drm: reject malformed headers earlyMira Okafor39m ago8bc4afe
sched: document the invariantMira Okafor48m ago0c8a05f
io_uring: drop an unused helperTomás Vega3h ago0f40e76
mm: defer the flush until commitPriya Nair4h ago11f7c8dChecks passed
Merge pull request #9793 from net/next-1Jonas Lindqvist4h ago9957171Checks passed
io_uring: fold two passes into oneMira Okafor5h ago913272c
mm: make the timeout configurableMira Okafor5h ago93e9543
btrfs: defer the flush until commitMira Okafor6h ago96a035a
locking: split the init sequenceJonas Lindqvist6h ago14aeaa4
Merge pull request #9786 from sched/next-2Tomás Vega7h ago9ec4d9fChecks passed
io_uring: widen the counter to 64 bitsPriya Nair7h ago9c0df88
ext4: split the init sequenceMira Okafor8h ago17658bb
nvme: report the right error codeTomás Vega8h ago1a1c6d2
Merge pull request #9779 from locking/next-3Jonas Lindqvist9h agoa6e97e4Checks passed
kvm: split the init sequenceMira Okafor9h agoa17bbb6
io_uring: defer the flush until commitMira Okafor10h agoa4329cd
perf: reject malformed headers earlyPriya Nair10h ago1cd34e9
ext4: document the invariantJonas Lindqvist11h ago1f8a300
io_uring: cache the lookup resultMira Okafor11h ago2241117
Merge pull request #9772 from fs/next-4Mira Okafor12h agoac57412Checks passed
btrfs: make the timeout configurableTomás Vega12h agoa9a05fb
mm: fold two passes into oneTomás Vega13h ago24f7f2e
Merge pull request #9765 from tcp/next-5Jonas Lindqvist13h agob47be57Checks passed
bpf: make the timeout configurableMira Okafor14h agoaf0e229
fs: reject malformed headers earlyMira Okafor14h agob1c5040
tcp: use a per-CPU counterPriya Nair15h ago27aed45
fs: defer the flush until commitJonas Lindqvist15h ago2a65b5cChecks passed
Merge pull request #9758 from bpf/next-6Jonas Lindqvist16h agoc20e4caChecks passed
sched: free the buffer on the error pathMira Okafor16h agob732c6e
drm: fold two passes into oneMira Okafor17h agob9e9a85
tcp: rework the retry loopMira Okafor17h agobca089c
mm: widen the counter to 64 bitsMira Okafor18h agobf576b3
net: cache the lookup resultMira Okafor18h ago2d1c973
ext4: avoid a double lock on teardownTomás Vega19h ago2fd378a
net: handle the empty batchPriya Nair19h ago328a5a1
perf: drop an unused helperJonas Lindqvist20h ago35413b8
Merge pull request #9751 from crypto/next-7Mira Okafor2d agoca32f0fChecks passed
bpf: drop an unused helperTomás Vega2d agoc4c52e1
x86: defer the flush until commitTomás Vega2d agoc77c0f8
nvme: widen the counter to 64 bitsMira Okafor2d ago37f81cf
x86: make the timeout configurableTomás Vega2d ago3aaefe6
Merge pull request #9744 from locking/next-8Jonas Lindqvist2d agod257954Checks passed
rcu: report the right error codeMira Okafor2d agocce9d26
net: drop an unused helperMira Okafor2d agocfa0b3d
io_uring: bound the queue under pressurePriya Nair2d ago3d65dfd
Merge pull request #9737 from nvme/next-9Priya Nair2d agoda7c399Checks passed
bpf: stop polling when idleJonas Lindqvist2d agod50e76b
nvme: use a per-CPU counterJonas Lindqvist2d agod7c5582
btrfs: fold two passes into oneJonas Lindqvist2d ago401cc14
arm64: simplify the fast pathMira Okafor2d ago42d3a2bChecks passed
Merge pull request #9730 from crypto/next-10Jonas Lindqvist2d agodfe9fc7Checks passed
crypto: cache the lookup resultMira Okafor2d agodd331b0
block: handle the empty batchTomás Vega2d ago458a842
io_uring: defer the flush until commitPriya Nair3d ago4841659

Priya Nair priya@northpier.example

3m ago · master

drm: widen the counter to 64 bits

No pull request, checks or deployments

试试看

  1. 用最快的速度滚动历史记录。

  2. 把滚动条直接拖到底部再拖回来。

  3. 选中列表深处的一条提交。

  • 首屏优先呈现

    gitā 会先读取最新的 500 条提交并立即显示,然后在后台继续读取。你永远不必等待整个历史加载完成。

  • 随滚动按需加载

    当你向下浏览列表时,后续提交会以每次 2,000 条的速度持续载入。

  • 始终保持可读的图谱

    无论仓库有多少分支,图谱横向最多绘制四十条泳道,并会提示其余分支的数量。

  • 为慢速仓库提速

    开启该设置后,gitā 会为读取较慢的仓库生成 git 的 commit-graph 文件并保持索引新鲜。你的提交与文件本身不会被改动。

  • 后台静默低耗

    关闭窗口后,gitā 仍在检查 GitHub、Vercel 和智能体状态,但仅占用不到单个 CPU 核心 1% 的资源。

  • 无网页视图

    列表、树形结构和差异对比均采用原生 Mac 视图,你与滚动条之间没有任何浏览器引擎拖慢速度。

Questions, *answered*

How does gitā stay fast with huge histories?
It asks git only for what it is about to show: the newest 500 commits come first, and further commits arrive 2,000 at a time as you scroll. The lists, trees and diffs are native Mac views.
Can gitā really open the Linux kernel?
Yes. The Linux kernel has about 1.5 million commits, and it opens and scrolls in gitā like any other repository.

继续探索

全部功能

下载 gitā

免费。支持 macOS 14 或更高版本,兼容 Apple silicon 与 Intel。