ac-native: wait up to 10s for GPU device (Gemini Lake is slow)

Gemini Lake (Yoga 11e) i915 probe takes longer than 1s.
Also check card1 in addition to card0.

Co-Authored-By: Claude Opus 4.6 (1M context) <[e-mail]>

ants/mail-mcp/server.mjs

···
 import { z } from "zod";
 import { execFile as execFileCb, spawn as spawnCb } from "node:child_process";
 import { promisify } from "node:util";
-import { access, readFile } from "node:fs/promises";
+import { access, readFile, stat } from "node:fs/promises";
+import { basename } from "node:path";
 
 const execFile = promisify(execFileCb);
 
···
 // --- mail_send ---
 server.tool(
   "mail_send",
-  "Send an email via msmtp. Defaults to [e-mail]; use from_account to switch. For replies, set in_reply_to and references for proper threading.",
+  "Send an email via msmtp. Defaults to [e-mail]; use from_account to switch. For replies, set in_reply_to and references for proper threading. Supports file attachments via the attachments parameter (array of absolute paths).",
   {
     to: z.string().describe("Recipient email address"),
     cc: z.string().optional().describe("CC email address(es), comma-separated"),
···
       .optional()
       .default("ac-mail")
       .describe("Which msmtp account to send from (ac-mail or jas-mail)"),
+    attachments: z
+      .array(z.string())
+      .optional()
+      .describe("Array of absolute file paths to attach (e.g. [\"/path/to/file.pdf\"])"),
   },
-  async ({ to, cc, subject, body, in_reply_to, references, preserve_case, signature, from_account }) => {
+  async ({ to, cc, subject, body, in_reply_to, references, preserve_case, signature, from_account, attachments }) => {
     try {
       const style = await loadEmailStyle();
       const styled = applyEmailStyle({
···
           .join(" ");
         headers.push(`References: ${refs}`);
       }
-      headers.push(
-        `Subject: ${styled.subject}`,
-        "MIME-Version: 1.0",
-        "Content-Type: text/plain; charset=UTF-8",
-        "Content-Transfer-Encoding: 8bit",
-      );
-      const message = [...headers, "", styled.body].join("\r\n");
+      headers.push(`Subject: ${styled.subject}`, "MIME-Version: 1.0");
+
+      let message;
+      if (attachments && attachments.length > 0) {
+        // MIME multipart with attachments
+        const boundary = `----=_Part_${Date.now()}_${Math.random().toString(36).slice(2)}`;
+        headers.push(
+          `Content-Type: multipart/mixed; boundary="${boundary}"`,
+        );
+        const parts = [
+          ...headers,
+          "",
+          `--${boundary}`,
+          "Content-Type: text/plain; charset=UTF-8",
+          "Content-Transfer-Encoding: 8bit",
+          "",
+          styled.body,
+        ];
+        for (const filePath of attachments) {
+          await stat(filePath); // throws if missing
+          const fileData = await readFile(filePath);
+          const name = basename(filePath);
+          const ext = name.split(".").pop()?.toLowerCase();
+          const mime = ext === "pdf" ? "application/pdf"
+            : ext === "png" ? "image/png"
+            : ext === "jpg" || ext === "jpeg" ? "image/jpeg"
+            : ext === "html" ? "text/html"
+            : ext === "xlsx" ? "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+            : "application/octet-stream";
+          parts.push(
+            `--${boundary}`,
+            `Content-Type: ${mime}; name="${name}"`,
+            "Content-Transfer-Encoding: base64",
+            `Content-Disposition: attachment; filename="${name}"`,
+            "",
+            fileData.toString("base64").replace(/(.{76})/g, "$1\r\n"),
+          );
+        }
+        parts.push(`--${boundary}--`);
+        message = parts.join("\r\n");
+      } else {
+        // Plain text (no attachments)
+        headers.push(
+          "Content-Type: text/plain; charset=UTF-8",
+          "Content-Transfer-Encoding: 8bit",
+        );
+        message = [...headers, "", styled.body].join("\r\n");
+      }
 
       const { stdout, stderr } = await run(MSMTP, ["-a", account, "-t"], { input: message });
       return {

fedac/native/src/ac-native.c

···
     // Bring up loopback interface (needed for Claude OAuth callback server)
     system("/bin/ip link set lo up 2>/dev/null || /usr/sbin/ip link set lo up 2>/dev/null || ifconfig lo up 2>/dev/null");
 
-    // Wait for display device (up to 1s)
-    for (int i = 0; i < 100; i++) {
+    // Wait for display device (up to 10s — Gemini Lake GPUs can be slow)
+    for (int i = 0; i < 1000; i++) {
         if (access("/dev/dri/card0", F_OK) == 0 ||
+            access("/dev/dri/card1", F_OK) == 0 ||
             access("/dev/fb0", F_OK) == 0) break;
         usleep(10000);
     }

microvision/blind-valuation-comparison.tex

···
 \item Grok specifically modeled the ``short killer'' dividend scenario; Claude focused on sum-of-parts
 \end{itemize}
 
+\section{Deep Dive: Supporting Market Data}
+
+The following data was gathered from current analyst reports, industry sources, and public filings to ground both AI valuations in real numbers.
+
+\subsection{Automotive LiDAR --- Verified Numbers}
+
+\begin{itemize}
+\item The global automotive LiDAR market reached \textbf{\$1.19B in 2024} and is projected to reach \textbf{\$9.59B by 2030} at a \textbf{41.6\% CAGR} (MarketsandMarkets)
+\item Alternative estimates: \$1.23B (2025) $\rightarrow$ \$5.31B (2030) at 34\% CAGR (Mordor Intelligence)
+\item Solid-state LiDAR at \textasciitilde\$200--250/unit is the inflection point where ADAS integration becomes cost-viable for mid-range vehicles, not just luxury (IEEE Spectrum, Feb 2026)
+\item Hesai announced at CES 2026 it will \textbf{double production capacity} from 2M to 4M+ units/year. ATX orders exceed 4M units. 24 automotive OEM design wins.
+\item ADAS applications now represent 38\% of total LiDAR demand, expected to surpass 60\% by 2030
+\end{itemize}
+
+\subsection{Current LiDAR Comparable Valuations (March 2026)}
+
+\begin{center}
+\begin{tabular}{@{}llr@{}}
+\toprule
+\textbf{Company} & \textbf{Focus} & \textbf{Market Cap} \\
+\midrule
+Hesai Group (HSAI) & Auto/Industrial LiDAR & \$3.59B \\
+Valeo & Auto LiDAR + ADAS & \$2.75B \\
+Ouster (OUST) & Multi-sector LiDAR & \$1.25B \\
+Aeva Technologies & Automotive FMCW LiDAR & \$0.84B \\
+indie Semiconductor & Auto sensing chips & \$0.53B \\
+Innoviz (INVZ) & Automotive LiDAR & \$0.14B \\
+AEye & Adaptive LiDAR & \$0.10B \\
+\midrule
+\textbf{Total LiDAR sector} & & \textbf{\$9.98B} \\
+\bottomrule
+\end{tabular}
+\end{center}
+
+{\small\textit{Source: companiesmarketcap.com, retrieved March 2026.}
+
+Key observation: A company with proven mass-production at \$250/unit \textit{plus} NED/AR capabilities would be the only LiDAR company spanning both sensing and display markets. Hesai at \$3.59B does LiDAR only. Adding NED IP and the AR glasses opportunity on top of a comparable LiDAR product line implies a valuation well above Hesai.}
+
+\subsection{Industrial: Forklift Safety \& Automation}
+
+\begin{itemize}
+\item \textbf{Autonomous forklift market:} \$7.84B (2025) $\rightarrow$ \$13.79B by 2030 at 12\% CAGR (GlobeNewsWire, Mar 2026)
+\item Alternative: \$5.75B (2025) $\rightarrow$ \$10.24B by 2030 at 12.25\% CAGR (Grand View Research)
+\item Overall forklift market: \$91.19B (2025) $\rightarrow$ \$141.32B by 2032 at 6.5\% CAGR
+\item \textbf{OSHA data:} 85 forklift fatalities, 34,900 serious injuries, 61,800 non-serious incidents per year in the US alone
+\item Forklift accidents cost an estimated \textbf{\$3.7 billion/year} in the US --- bolt-on LiDAR safety directly reduces this liability
+\item LiDAR-based SLAM systems reduce pick-and-place errors by \textbf{up to 70\%} and increase pallet productivity by \textbf{25\%}
+\item Crown, Hyster, and Yale all now offer LiDAR-based proximity systems --- proving market demand
+\item A bolt-on retrofit that integrates with existing drive systems avoids the \$30K--100K cost of replacing an entire forklift
+\end{itemize}
+
+\subsection{Heavy Equipment \& Defense Automation}
+
+\begin{itemize}
+\item \textbf{Autonomous construction equipment:} \$4.56B (2024) $\rightarrow$ \$9.74B by 2030 at 13.5\% CAGR (MarketsandMarkets)
+\item Remote Operation Centers allow operators to \textbf{control multiple fleets across continents} --- LiDAR is the critical sensing layer
+\item By 2026, \textbf{over 30\% of global mining operations} are projected to use fully autonomous haulage systems
+\item By 2026, autonomous mining fleets expected to operate at \textbf{up to 85\% full automation}
+\item Caterpillar, Komatsu, John Deere, Oshkosh all showcased autonomous equipment at CES 2025
+\item Applications radiating into agriculture, forestry, infrastructure, and defense
+\end{itemize}
+
+\subsection{Near-Eye Display \& Military AR}
+
+\begin{itemize}
+\item \textbf{NED market revenue:} \$1.2B in 2026 (first time surpassing \$1B), over \textbf{200\% YoY growth}
+\item Projected to reach \$2.8B by 2028 and \textbf{\$6.65B--9.55B by 2030} (25\% CAGR)
+\item NED shipments: 20.2M units (2026) $\rightarrow$ \textbf{92.9M units by 2030}
+\item AI glasses shipments forecast to exceed \textbf{10M units by 2026}
+\item Meta, Apple, Huawei, Pico all launching new MR products in 2026--27
+\item The broader AR hardware market crossed \textbf{\$80B in 2024}, with 37.9\% CAGR forecasts to \$1.7T by 2032
+
+\item \textbf{Military:} The US Army's IVAS program is a \textbf{\$21.88 billion} contract for 120,000+ AR headsets. Originally Microsoft; contract novated to \textbf{Anduril} in April 2025. MEMS laser beam scanning is a core enabling technology for the display.
+\item DARPA awarded \$4.7M specifically to transform AR glasses technology
+\item Syntec Optics received defense orders for AR display components readable in day/night combat
+
+\item \textbf{NED IP deal comps:}
+\begin{itemize}
+\item Microsoft / Osterhout Group AR patents: \textasciitilde\$150M (2014)
+\item Google / HTC XR IP license: \$250M (2025)
+\item Infinite Reality / Touchcast AR/AI: \$500M (2025)
+\item Meta / CTRL-Labs neural interface: \$1B (2019)
+\item Apple acquired 4 AR companies (2015--16): Flyby Media, metaio, Emotient, Faceshift
+\item Apple filed \textbf{5,000 patents} for Vision Pro technology alone
+\end{itemize}
+
+\item A proven, mass-producible miniature laser projector enabling HD 3D in Ray Ban-sized frames is \textbf{the missing component} that Meta, Apple, and EssilorLuxottica (market cap: 126.5B EUR) need for consumer AR glasses
+\end{itemize}
+
+\section{Bottom Line}
+
+\fcolorbox{acpink}{acpink!6}{\parbox{\dimexpr\textwidth-2\fboxsep-2\fboxrule}{%
+\textbf{\color{acpink}Both AI engines, working blind, independently concluded:}
+
+\begin{enumerate}
+\item A company with these capabilities should have a market capitalization of \textbf{\$2--7 billion} (base case), with upside to \$10B+
+\item The Near-Eye Display IP alone is worth \textbf{\$200--600 million} --- a licensing deal or sale at this level would be transformative
+\item The \textbf{combination} of automotive LiDAR + industrial bolt-on + heavy equipment automation + AR/NED in a single platform is unique --- no pure-play competitor spans all four markets
+\item The industrial and defense segments are \textbf{nearer-term revenue drivers} than automotive, where OEM cycles are longer
+\item The military AR opportunity (\$21.88B IVAS alone) and consumer AR inflection (92.9M NED units by 2030) provide massive tailwinds
+\end{enumerate}
+
+These valuations are grounded in verified TAM data from MarketsandMarkets, Grand View Research, Mordor Intelligence, Omdia, OSHA, and public comparable company market caps as of March 2026.}}
+
 \vspace{1em}
 \hrule
 \vspace{0.5em}

vscode-extension/package-lock.json

···
 {
   "name": "aesthetic-computer-code",
-  "version": "1.269.0",
+  "version": "1.268.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "aesthetic-computer-code",
-      "version": "1.269.0",
+      "version": "1.268.0",
       "license": "None",
       "dependencies": {
         "acorn": "^8.15.0",