Web and desktop editions ======================== Both editions share the tabs, fit engine, model library, parameter plots, and configuration-sharing protocol. The differences come from where the data and the command-line tools live. .. list-table:: :header-rows: 1 :widths: 25 35 40 * - Concern - Desktop (Electron) - Web (browser) * - Where the tools run - Locally using configured or detected tools; through WSL2 on Windows. - On the server behind the CGI endpoint. * - Data access - Local data plus optional downloads of missing selected runs. - What the server data tree exposes. * - Preferences - Per-user, through **File / Preferences**. - Fixed by server configuration; no dialog. * - Opening / saving MSR files - Native **File** menu and editor Save button. - Server-side artifacts; no local file dialog. * - Spectra display - ``musrview`` in its own native window. - ROOT canvas produced by ``musrview`` on the server, drawn with JSROOT in the page. * - Quick Start help - Independent auxiliary window. - Modeless dialog beside the main window. * - Configuration links - Share webpage links; **Restore configuration link** to paste one. - Share button copies a link; opening it restores the settings. Where the runtime comes from ---------------------------- Windows packages include a WSL2 runtime named ``muSRFlow-runtime``. Linux packages keep the frontend and scientific runtime separate, allowing either a local ROOT/musrfit installation or a published ``musrflow-runtime`` package. The web server owns its runtime and data paths. See :doc:`installation` for the package choices instead of configuring these components manually. The workflow chapters of this manual apply to both editions. Where a step is specific to one edition (handling files, Preferences), the text says so explicitly. Downloading missing runs ------------------------ Fit and Plot check selected main runs and ADDRUN members before preparing MSR files. If files are missing, the desktop offers to download them over HTTPS from ``musruser.psi.ch`` and continue. Independent series fits request data only for entries that need fitting; reused results and explicit skips do not trigger downloads. Runs absent from the server are still reported as missing. The writable Data directory is preferred. When it is read-only, a persistent folder under the user's application data directory is used instead. Exact filenames and formats are remembered across sessions, and incomplete downloads are never published as run files. Expert MSRs with manually specified RUN paths continue to use their existing file-resolution rules. Generated download RUN paths are relative to the stored data tree. muSRFlow passes the matching search roots to its tools through ``MUSRFULLDATAPATH``, including translated paths for Windows/WSL2. To run such an MSR from a terminal outside muSRFlow, set that variable to the download root or add the root to your own musrfit startup XML.