<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Eric&apos;s Blog</title><description>Research notes, data wrangling, and a few small tools.</description><link>https://erictseng.icu/en/</link><language>en</language><item><title>Pitfalls I hit on MySQL 5.7</title><link>https://erictseng.icu/en/blog/mysql57-pitfalls/</link><guid isPermaLink="true">https://erictseng.icu/en/blog/mysql57-pitfalls/</guid><description>Every four-byte character compares equal, no CTEs, half-dead connections, and a slow public endpoint.</description><pubDate>Sun, 27 Sep 2026 00:00:00 GMT</pubDate></item><item><title>Why OCR can&apos;t fix rare Chinese characters</title><link>https://erictseng.icu/en/blog/ocr-rare-characters/</link><guid isPermaLink="true">https://erictseng.icu/en/blog/ocr-rare-characters/</guid><description>Three root causes, four dead ends, and a single-character classifier trained in 40 minutes.</description><pubDate>Sun, 27 Sep 2026 00:00:00 GMT</pubDate></item><item><title>How many dpi does book scanning really need?</title><link>https://erictseng.icu/en/blog/scan-dpi/</link><guid isPermaLink="true">https://erictseng.icu/en/blog/scan-dpi/</guid><description>The quotes for 300 and 600 dpi are far apart. I did the math with my own recognition pipeline and existing scans.</description><pubDate>Sun, 27 Sep 2026 00:00:00 GMT</pubDate></item></channel></rss>