Back to journal

One source for a bilingual journal

On August 28 and 29, I simplified the way bogdanalin.com turns journal entries into public pages. Each entry already needed to appear in several places: an English page, a Romanian page, the archives, feeds and sitemap. Keeping those derived files alongside the authored text made it harder to tell which changes belonged to the article and which were consequences of rebuilding the site.

The completed change made the dated JSON entry the canonical source. The normal build reads that source and produces the journal pages and their related output in the deployment directory. Generated journal files no longer need to be edited or committed as a parallel version of the same article. The English and Romanian narratives remain together in the source, where their dates, links and meaning can be reviewed together.

Removing tracked output was only useful if the resulting website stayed equivalent. I added and ran checks around fresh output, repeatable generation and the separation between authored files and build products. The final review also tightened the boundaries around where output could be written. The dated acceptance records cover focused checks and the full local verification gate; they do not stand in for a production observation.

The change stayed specific to the journal. Existing project-detail pages retained their own generation workflow. That distinction matters because a cleanup can become a migration of unrelated systems if a shared word such as generated is treated as sufficient reason to change everything. Here, the smaller boundary kept the source model clear without forcing other content through it.

Reflection

For me, the useful simplification was not just fewer tracked files. It was a clearer answer to where an article lives and how its public forms are produced. A single reviewed source plus reproducible output is easier to maintain than several editable copies that must be kept in agreement.