<?xml version="1.0"?>
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
	<id>https://txwiki.com/api.php?action=feedcontributions&amp;feedformat=atom&amp;user=IrisField</id>
	<title>Txwiki - User contributions [en]</title>
	<link rel="self" type="application/atom+xml" href="https://txwiki.com/api.php?action=feedcontributions&amp;feedformat=atom&amp;user=IrisField"/>
	<link rel="alternate" type="text/html" href="https://txwiki.com/wiki/Special:Contributions/IrisField"/>
	<updated>2026-11-03T02:01:56Z</updated>
	<subtitle>User contributions</subtitle>
	<generator>MediaWiki 1.43.9</generator>
	<entry>
		<id>https://txwiki.com/index.php?title=%E7%AB%99%E7%82%B9%E8%B4%A6%E5%8F%B7%E8%BD%AE%E6%8D%A2%E7%AD%96%E7%95%A5%E8%AF%B4%E6%98%8E&amp;diff=12</id>
		<title>站点账号轮换策略说明</title>
		<link rel="alternate" type="text/html" href="https://txwiki.com/index.php?title=%E7%AB%99%E7%82%B9%E8%B4%A6%E5%8F%B7%E8%BD%AE%E6%8D%A2%E7%AD%96%E7%95%A5%E8%AF%B4%E6%98%8E&amp;diff=12"/>
		<updated>2026-10-29T08:37:14Z</updated>

		<summary type="html">&lt;p&gt;IrisField: [ProfileMax] 发布测试&lt;/p&gt;
&lt;hr /&gt;
&lt;div&gt;== 为什么要轮换站点账号 ==&lt;br /&gt;
&lt;br /&gt;
同一账号在高频发布节奏下容易被站点风控标记,轻则限速、重则封禁。把发布压力均匀摊到站点下的多个账号上,是保持长期可用性的基础手段。&lt;br /&gt;
&lt;br /&gt;
== 轮换策略 ==&lt;br /&gt;
&lt;br /&gt;
* &#039;&#039;&#039;最近最久未用&#039;&#039;&#039;:每次发布优先选择 lastLoginAt 最靠前的可用账号;&lt;br /&gt;
* &#039;&#039;&#039;失败退避&#039;&#039;&#039;:登录失败或写入被拒的账号直接进入冷却,下一次投递自动换人;&lt;br /&gt;
* &#039;&#039;&#039;凭据分层&#039;&#039;&#039;:active 状态且密码齐全的账号才进入候选池,无凭据账号会被健康检查标 expired。&lt;br /&gt;
&lt;br /&gt;
== 操作约定 ==&lt;br /&gt;
&lt;br /&gt;
批量发布作业按站点串行投递,同一站点同一时刻只占用一个浏览器身份(Profile),发布成功后回写账号最近使用时间,保证轮换计量准确。&lt;br /&gt;
&lt;br /&gt;
轮换器的实现与发布测试共享同一套账号池,压测示范站点为 [https://txwiki.com/wiki/Main_Page Txwiki]。&lt;br /&gt;
&lt;br /&gt;
&amp;lt;blockquote&amp;gt;账号池告警阈值:单站点可用账号少于 2 个时,应导入新凭据而不是压缩发布节奏。&amp;lt;/blockquote&amp;gt;&lt;/div&gt;</summary>
		<author><name>IrisField</name></author>
	</entry>
	<entry>
		<id>https://txwiki.com/index.php?title=%E5%8F%91%E5%B8%83%E8%B4%A8%E9%87%8F%E6%A3%80%E6%9F%A5%E6%B8%85%E5%8D%95&amp;diff=11</id>
		<title>发布质量检查清单</title>
		<link rel="alternate" type="text/html" href="https://txwiki.com/index.php?title=%E5%8F%91%E5%B8%83%E8%B4%A8%E9%87%8F%E6%A3%80%E6%9F%A5%E6%B8%85%E5%8D%95&amp;diff=11"/>
		<updated>2026-10-29T08:34:03Z</updated>

		<summary type="html">&lt;p&gt;IrisField: [ProfileMax] 发布测试&lt;/p&gt;
&lt;hr /&gt;
&lt;div&gt;== 发布后质量检查清单 ==&lt;br /&gt;
&lt;br /&gt;
每一篇发布成功的文章都要过以下检查,任一项不过都应进入修复流程。&lt;br /&gt;
&lt;br /&gt;
=== 基础可达 ===&lt;br /&gt;
&lt;br /&gt;
* 发布页 HTTP 状态码 200(重定向后可达);&lt;br /&gt;
* 页面标题与文章标题一致。&lt;br /&gt;
&lt;br /&gt;
=== 链接与 rel ===&lt;br /&gt;
&lt;br /&gt;
* 目标链接在页面 HTML 中存在且 href 与投放计划一致;&lt;br /&gt;
* rel 属性按投放计划携带(nofollow/sponsored 等),缺失记 missing;&lt;br /&gt;
* 无死链(跳转后 404 即 dead)。&lt;br /&gt;
&lt;br /&gt;
=== 排版与内容 ===&lt;br /&gt;
&lt;br /&gt;
* HTML→wikitext 转换无内容丢失(标题/列表/加粗原样保留);&lt;br /&gt;
* 段落无明显截断,末尾完整。&lt;br /&gt;
&lt;br /&gt;
示范站点 [https://txwiki.com/wiki/Main_Page Txwiki] 的检查记录可在「发布 → 质量评分」Tab 查询。&lt;/div&gt;</summary>
		<author><name>IrisField</name></author>
	</entry>
	<entry>
		<id>https://txwiki.com/index.php?title=%E7%AB%99%E7%82%B9%E8%B4%A6%E5%8F%B7%E8%BD%AE%E6%8D%A2%E7%AD%96%E7%95%A5%E8%AF%B4%E6%98%8E&amp;diff=10</id>
		<title>站点账号轮换策略说明</title>
		<link rel="alternate" type="text/html" href="https://txwiki.com/index.php?title=%E7%AB%99%E7%82%B9%E8%B4%A6%E5%8F%B7%E8%BD%AE%E6%8D%A2%E7%AD%96%E7%95%A5%E8%AF%B4%E6%98%8E&amp;diff=10"/>
		<updated>2026-10-29T08:31:39Z</updated>

		<summary type="html">&lt;p&gt;IrisField: [ProfileMax] 发布测试&lt;/p&gt;
&lt;hr /&gt;
&lt;div&gt;== 为什么要轮换站点账号 ==&lt;br /&gt;
&lt;br /&gt;
同一账号在高频发布节奏下容易被站点风控标记,轻则限速、重则封禁。把发布压力均匀摊到站点下的多个账号上,是保持长期可用性的基础手段。&lt;br /&gt;
&lt;br /&gt;
== 轮换策略 ==&lt;br /&gt;
&lt;br /&gt;
* &#039;&#039;&#039;最近最久未用&#039;&#039;&#039;:每次发布优先选择 lastLoginAt 最靠前的可用账号;&lt;br /&gt;
* &#039;&#039;&#039;失败退避&#039;&#039;&#039;:登录失败或写入被拒的账号直接进入冷却,下一次投递自动换人;&lt;br /&gt;
* &#039;&#039;&#039;凭据分层&#039;&#039;&#039;:active 状态且密码齐全的账号才进入候选池,无凭据账号会被健康检查标 expired。&lt;br /&gt;
&lt;br /&gt;
== 操作约定 ==&lt;br /&gt;
&lt;br /&gt;
批量发布作业按站点串行投递,同一站点同一时刻只占用一个浏览器身份(Profile),发布成功后回写账号最近使用时间,保证轮换计量准确。&lt;br /&gt;
&lt;br /&gt;
&amp;lt;blockquote&amp;gt;账号池告警阈值:单站点可用账号少于 2 个时,应导入新凭据而不是压缩发布节奏。&amp;lt;/blockquote&amp;gt;&lt;/div&gt;</summary>
		<author><name>IrisField</name></author>
	</entry>
	<entry>
		<id>https://txwiki.com/index.php?title=MediaWiki_%E6%89%B9%E9%87%8F%E5%8F%91%E5%B8%83%E6%8C%87%E5%8D%97(%E5%8E%8B%E6%B5%8B%E7%89%88)&amp;diff=9</id>
		<title>MediaWiki 批量发布指南(压测版)</title>
		<link rel="alternate" type="text/html" href="https://txwiki.com/index.php?title=MediaWiki_%E6%89%B9%E9%87%8F%E5%8F%91%E5%B8%83%E6%8C%87%E5%8D%97(%E5%8E%8B%E6%B5%8B%E7%89%88)&amp;diff=9"/>
		<updated>2026-10-29T08:27:43Z</updated>

		<summary type="html">&lt;p&gt;IrisField: [ProfileMax] 发布测试&lt;/p&gt;
&lt;hr /&gt;
&lt;div&gt;== 为什么做批量发布压测 ==&lt;br /&gt;
&lt;br /&gt;
批量发布的价值不只在吞吐量,更在于把「发布成功率」从主观印象变成可验证的指标:站点可用性、凭据有效性、落栈结果、以及发布后的回读校验,全部由系统记录并汇总成质量评分。&lt;br /&gt;
&lt;br /&gt;
本轮压测使用 15 个 MediaWiki 压测站点,每个站点绑定一个专属发布镜像(指纹浏览器 Profile)与站点账号池,发布作业由调度器按站点串行投递,避免同账号并发发布触发风控。&lt;br /&gt;
&lt;br /&gt;
== 账号导入与凭据管理 ==&lt;br /&gt;
&lt;br /&gt;
* 站点账号密码在导入时即以 AES-256-GCM 加密落库,API 永不返回明文,只在投递时解密注入执行载荷;&lt;br /&gt;
* 每个站点保留原始创建时间与最近登录时间,账号轮换按「最近最久未用」进行;&lt;br /&gt;
* 无有效凭据的账号会被健康检查标记为 expired,不再进入投递候选。&lt;br /&gt;
&lt;br /&gt;
== 发布流程与回读校验 ==&lt;br /&gt;
&lt;br /&gt;
发布作业分两种模式:&#039;&#039;&#039;publish&#039;&#039;&#039; 直接落正式页面,&#039;&#039;&#039;draft&#039;&#039;&#039; 落到账号个人空间的沙盒页,用于在不动正式内容的前提下验证登录与写入链路。落栈成功后系统记录 revisionId,并用 HTTP 回读发布页:确认页面可访问(HTTP 200)、目标链接存在且 rel 属性符合预期。&lt;br /&gt;
&lt;br /&gt;
&amp;lt;blockquote&amp;gt;发布结果(status/message/revisionId/articleUrl)与目标链接检查记录都会在「发布任务 / 质量评分」Tab 中按站点汇总,压测结论以这两处数据为准。&amp;lt;/blockquote&amp;gt;&lt;br /&gt;
&lt;br /&gt;
== 压测站点附录 ==&lt;br /&gt;
&lt;br /&gt;
本轮参与压测的站点包括 youngwiki、eastwiki、hopewiki 等 15 个 MediaWiki 环境,完整站点清单见「发布 → 站点」Tab;示范站点为 [https://txwiki.com/wiki/Main_Page 压测站点 Txwiki]。&lt;/div&gt;</summary>
		<author><name>IrisField</name></author>
	</entry>
	<entry>
		<id>https://txwiki.com/index.php?title=Microsoft%27s_Reported_Plan_to_Buy_Out_a_Quarter_of_Its_Employees,_Explained&amp;diff=8</id>
		<title>Microsoft&#039;s Reported Plan to Buy Out a Quarter of Its Employees, Explained</title>
		<link rel="alternate" type="text/html" href="https://txwiki.com/index.php?title=Microsoft%27s_Reported_Plan_to_Buy_Out_a_Quarter_of_Its_Employees,_Explained&amp;diff=8"/>
		<updated>2026-10-28T19:48:50Z</updated>

		<summary type="html">&lt;p&gt;IrisField: Reports circulating in the technology press describe Microsoft preparing voluntary buyout offers that could extend to a substantial share of its workforce — by some accounts, roughly a quarter of the company&amp;#039;s employees. Employees in the United State&lt;/p&gt;
&lt;hr /&gt;
&lt;div&gt;Reports circulating in the technology press describe Microsoft preparing voluntary buyout offers that could extend to a substantial share of its workforce — by some accounts, roughly a quarter of the company&#039;s employees. Employees in the United States, Canada, India, and the United Kingdom are said to be among those who may receive offers to leave the company in exchange for a separation package. For readers who have encountered the word &amp;quot;buyout&amp;quot; mainly in headlines, the terminology can be confusing: in this context a buyout has nothing to do with acquiring a company, and everything to do with paying employees to leave voluntarily. This article explains what is reportedly happening at Microsoft, defines the core concepts behind such programs, and walks through how they typically operate.&lt;br /&gt;
&lt;br /&gt;
== What Is Happening at Microsoft ==&lt;br /&gt;
&lt;br /&gt;
According to recent reports, Microsoft has been preparing a voluntary separation program under which employees in the US, Canada, India, and the UK may receive offers to leave the company. Figures cited in reports suggest the potential scope could reach approximately one in four employees — a substantial share of the workforce for a program of this kind.&lt;br /&gt;
&lt;br /&gt;
Several points are worth emphasizing. Programs like this usually apply to particular business units, job categories, or geographic regions rather than to an entire workforce simultaneously, so the headline figure and the number of people actually receiving offers may differ. The voluntary nature also matters: the final effect depends on how many employees choose to accept. As of the available reporting, specifics such as which teams are affected, the exact value of individual packages, and the expected take-up rate have not been fully detailed.&lt;br /&gt;
&lt;br /&gt;
What is clear is the mechanism. Rather than unilaterally cutting roles, the company would invite eligible employees to depart on negotiated terms&lt;br /&gt;
&lt;br /&gt;
== References ==&lt;br /&gt;
* [https://arstechnica.com/gadgets/2026/10/microsoft-offers-voluntary-buyouts-to-60000-employees/ Microsoft offers to buy out a quarter of its workforce]&lt;br /&gt;
&lt;br /&gt;
[[Category:microsoft]]&lt;br /&gt;
[[Category:employees]]&lt;br /&gt;
[[Category:us]]&lt;br /&gt;
[[Category:canada]]&lt;br /&gt;
[[Category:india]]&lt;/div&gt;</summary>
		<author><name>IrisField</name></author>
	</entry>
</feed>