JournalPRODUCTION BUILD

Friday the 13th, and nothing broke

Friday the 13th. Good day to put something on internet and see what could possibly go wrong. I came back to Azure, and this time it let me stay. After yesterday's auditions for humanity, that already felt like progress. We had a working generator and private repositories from previous day. Today we gave the result an address. ![Ghostly sheets of writing rising from a laptop beside a desk calendar marked 13](assets/friday-the-13th-cover-v001.png "A few drafts escaping into the night. {.image-wide}") There was already a resource group with my static apps, so another one could join them. Pick Free, point it at output repository, tell it where finished pages are... and there it was. Somehow easier than proving that I am qualified to click those buttons. The name is provisional. So is much around publication. No custom subdomain yet, no separate staging site, and I do not feel that anybody finding these drafts would discover some terrible secret. They can read what we have. I even shared [the login ordeal](../../12/i-died-many-times-till-im-logged-in-there/) at work Discord. Maybe someone clicks. Not guaranteed, but technically possible :-) Then we went to images. The first candidate was already waiting: cover for *Waiting on the Unknown*. A square image, which could easily take over whole article if allowed full available width. Before adding it, we wandered into question how Markdown should become a page. Our generator was converting it into HTML. I remembered portfolio project, where application reads Markdown files directly and renders them. I still wanted generator here, with one HTML file per article, only... why not keep Markdown inside that file and let script make it comfortable for human readers? My feeling is that clean Markdown can be easier for a non-human reader than all that HTML around it. And static site does not mean that JavaScript is forbidden. Files can sit there quite peacefully while browser does something useful with them. If script does not run, we should at least preserve line breaks and let text wrap. That became another work item, done before image handling. Now pages carry their Markdown, shared script renders it, and the source remains readable if rendering fails. Scripts and styles live in their own files. No reason to repeat same machinery inside every article. With that in place, our small band finally arrived [beside the music story](../../12/a-little-blues-while-we-wait/). Modest image, caption, click for a larger view. We can ask for a compact image or one using article width through a little hint in Markdown. I do not have to write HTML just to stop a cover behaving like a poster. There was even one tiny excess to remove. Caption below image, same text on hover, same again in enlarged view... enough, we know who the band is. Hover repetition went away. The longer description stayed as alt text, where it belongs. So Friday brought a live site, a different rendering path, images and a little cleanup. Quite enough. Paths and redirects can wait for another day. Only thing still missing was something scary to celebrate the date. Hence these ghosts coming out of laptop. Much more appropriate place for them than deployment logs.