.gitattributes

# Line-ending policy: opt in, never repo-wide.
#
# WHY THIS EXISTS
#
# Five files under services/field-ops/frontend/src were stored with CRLF while
# the thirty-odd beside them were LF — an accident of which editor touched each
# one first, not a decision. The cost only appears later: any tool that
# rewrites such a file converts it wholesale, and the 2,512-line diff that
# results hides the twenty lines that actually changed. That happened while
# adding the Today screen, and the diff had to be split into two commits to
# stay reviewable.
#
# WHY IT IS NOT `* text=auto`
#
# That was the first attempt and it is wrong for this repository. 200+ tracked
# files are legitimately CRLF and some of them care:
#
#   *.bat, *.cmd   legacy PHIVOLCS batch scripts — CRLF is the format
#   *.rtl          VADASE NMEA capture fixtures the replay tests read
#   *.PCF *.INP    Bernese BPE panels, fixed-column and read by Fortran
#   *.STA *.CRD    Bernese station/coordinate files, same
#   *.m            PHIVOLCS MATLAB, edited on Windows
#
# A blanket rule would rewrite any of those the next time someone edited and
# re-added it, silently, in a commit about something else. That is the exact
# failure this file is meant to prevent, pointed at the data instead of the
# source. So: no default. Everything keeps whatever it already has, and
# normalisation is listed explicitly per extension below.

# Source we maintain. LF in the repository and in the working tree; these are
# edited on Linux by both machines and by tooling that assumes LF.
*.ts   text eol=lf
*.tsx  text eol=lf
*.js   text eol=lf
*.jsx  text eol=lf
*.css  text eol=lf
*.html text eol=lf

# Shell scripts must be LF everywhere: a CR after a shebang makes the
# interpreter unfindable, and this repo hands scripts to the R740 and to gps3.
*.sh text eol=lf

# Deliberately absent: *.py. 82 tracked Python files are CRLF today and
# converting them belongs in its own change, verified against the test suite,
# not smuggled in beside a UI screen.

# Binary — never examined for line endings, never diffed as text.
*.png  binary
*.jpg  binary
*.jpeg binary
*.gif  binary
*.pdf  binary
*.docx binary
*.xlsx binary
*.zip  binary
*.gz   binary
*.crx  binary