/* ============================================================================ TFM v8 -- THE colour file. src/web/shared/tfm-tokens.css EVERY COLOUR IN THIS BUILD IS DEFINED HERE AND NOWHERE ELSE. `p8_map_gate.py` rule 1 fails on a colour literal found in any other .css, .js or .html we wrote. One file is what makes a palette changeable; a hex code in a page is a palette nobody can find. WHY THIS IS CSS AND NOT JAVASCRIPT, WHICH IS WHERE IT USED TO LIVE. The palette was previously VALUES in shared/tfm-colors.js, written onto :root at run time by applyTheme(). That made every colour on every page depend on a module executing. On 2026-08-08 a module stopped executing and the effect was measured live: the sign-in page rendered completely unstyled -- not degraded, unstyled, because the stylesheet held nothing but var() references and none of them resolved. A stylesheet cannot fail to execute. It has no imports to link and no run-time errors. Defining the tokens here means the worst a broken module can now do is break that module, which is the blast radius it should have had. The .js half is the READER, not a second source: shared/tfm-colors.js resolves these same custom properties for the code that needs a colour as a string (OpenLayers styles, a border set from script). It holds no literal. ========================================================================== */ :root { /* ---- the shell. v7's LIGHT ground and teal accent, carried across ------- **RULED 2026-09-01 (dashboard shape and look): v7's look on every page, including the map. ONE token set -- the theme is not forked, and there is no dark variant to drift.** These five and the six below them are the only values that moved; the twelve TIA-598-C strand colours did NOT, because those are the industry's and not ours to choose. What makes a pale strand readable here is the BORDER the swatch already carried (`.sv-swatch { border: 1px solid var(--tfm-edge) }`) -- a rule written for a light panel back when the panel was dark, and only now doing its job. */ --tfm-ground: #f6f8fa; /* page */ --tfm-panel: #ffffff; /* cards, toolbars, panels */ --tfm-panel-high: #f0f3f6; /* a raised row inside a panel */ --tfm-edge: #d8dee4; /* hairlines */ --tfm-ink: #1c2430; /* text */ /* **THE TWELVE TIA-598-C FIBRE COLOURS** (brief 7 section 2.3). Here because `p8_map_gate` refuses a colour literal anywhere else, and it is right to: a swatch hard-coded in a module is a colour the theme cannot reach. The ORDER and the NAMES are the database's -- `config.lookup_values` seeded them and `api.fn_fiber_color` serves them; these are only what each name looks like. **White and Slate carry a border** because a white dot on a light panel is an invisible dot. */ --tfm-fibre-blue: #1f6feb; --tfm-fibre-orange: #e8830c; --tfm-fibre-green: #2ea043; --tfm-fibre-brown: #8b5a2b; --tfm-fibre-slate: #8b98a5; --tfm-fibre-white: #f4f6f8; --tfm-fibre-red: #e5484d; --tfm-fibre-black: #24292f; --tfm-fibre-yellow: #e3b341; --tfm-fibre-violet: #a371f7; --tfm-fibre-rose: #f778ba; --tfm-fibre-aqua: #39c5cf; --tfm-bad: #c0271f; /* a located fault: the one red on the map */ --tfm-ink-muted: #5b6875; /* secondary text */ --tfm-accent: #127f75; /* TFM teal, darkened for a light ground -- same hue */ --tfm-accent-ink: #ffffff; /* text ON the accent */ --tfm-live: #127f75; /* the "live" tag */ --tfm-admin-tag: #9a6206; /* the "admin" tag */ /* ---- paper ------------------------------------------------------------- A printed report is not the screen. These are the only colours the print block uses, and they live here for the same reason every other colour does: a hex code in a page's @media print rule is a second palette that the one colour file cannot see and rule 1 would refuse. */ --tfm-paper: #ffffff; --tfm-paper-ink: #000000; --tfm-paper-edge: #999999; /* **THE FOUR SPLICE STATES** (brief 11, amendment 1 ruling 6). The status legend is MIRRORED from the QGIS splice window, and a legend is its labels AND the colours they explain -- so the tints stopped being hex literals in a Qt module the day the web had to show them. `core/spliceview.py` holds Qt's; these are the browser's, and `web_splice_parity_gate` binds the two to the same four NAMES -- **names, not hexes, which is what lets this surface pick its own values without breaking parity.** The hexes differ on purpose: these sit on this build's ground and the ones in the plugin sit on Qt's. **Lightened 2026-09-01** with the ruling: four dark tints meant to be read as fills BEHIND text, moved from a dark ground to a light one. */ --tfm-splice-free: #d5ecd9; --tfm-splice-circuit-gap: #f6e7c2; --tfm-splice-spliced: #e4e7ea; --tfm-splice-circuit-spliced: #f6d5d5; /* **THE TWO HIGHLIGHT SIDES** (brief 11, amendment 4 ruling 2). Picking a cable in the LEFT pane of the splice view and one in the RIGHT lights both on the map at once, in two colours, which is v7's own behaviour -- left blue, right yellow there. The same cable picked on both sides carries both, because v7 keys the feature by cable AND side. Here rather than in a module for the reason every colour is here: `p8_map_gate` rule 1 refuses a colour literal anywhere else, and a hex in a module is a colour the theme cannot reach. */ --tfm-hl-left: #2f6fed; --tfm-hl-right: #e3b341; /* The soft underlay beneath each splice curve. v7 draws `rgba(0,0,0,0.22)` -- 22% BLACK, under a fibre colour, on v7's WHITE sheet. **Copied literally onto this build's dark ground it is invisible**, which is half of why a port-to-port lane looked empty on 2026-08-20. The role is "separate the curve from the ground it crosses"; the value is this surface's own, which is the colour decision of amendment 1 section 9 applied where it was written to apply. */ --tfm-splice-lane-shadow: rgba(0, 0, 0, 0.22); /* **A SPLICE WITH NO FIBRE COLOUR STILL DRAWS A LINE** (amendment 6 ruling 2). A port has no colour, and the CELL rule is to leave an unknown colour unpainted rather than guess -- which is right for a swatch and wrong for a line, because the line IS the splice. v7 falls back to grey for exactly this. */ --tfm-splice-line-plain: #8b98a5; /* ---- the navigation rail (2026-09-01, look part two) -------------------- **A DARK SIDEBAR ON A LIGHT PAGE IS NOT A SECOND THEME.** It is one surface inside the one token set -- there is no `data-theme`, no media query and no second file, which is what "do not fork the theme" rules out. The dark ground did not leave the build; it stopped being the page and became the rail, which is v7's own arrangement. **THE NEW PAIR IS MEASURED, NOT ASSUMED** (part two: "a dark sidebar with light text is a new pair nothing has checked yet"). Ratios against `--tfm-nav` are in `src/AS-BUILT.md` under 2026-09-01 and are re-measurable from this file alone. */ --tfm-nav: #0f1722; /* the rail */ --tfm-nav-high: #1c2736; /* the current item, and hover */ --tfm-nav-ink: #e6edf6; /* a label */ --tfm-nav-ink-muted: #93a3b8; /* the section caption above the items */ --tfm-nav-edge: #212d3e; /* the rail's own hairline */ --tfm-nav-mark: #2fd4c4; /* the current item's left rule -- v7's teal, which reads on the rail where the page's darkened accent would not */ /* ---- states ------------------------------------------------------------ */ --tfm-warn: #b3261e; --tfm-ok: #1a7f37; --tfm-muted: #6a737d; --tfm-select: #0969da; /* ---- the ten served layers --------------------------------------------- */ /* One per layer in the tile index. A layer with no token here is drawn in --tfm-muted and says so in the console: an unknown layer given a colour of its own reads as a decision somebody made. */ --tfm-layer-conduits: #8c6d3f; --tfm-layer-fiber_cables: #1f6feb; --tfm-layer-splice_closures: #d29922; --tfm-layer-vaults: #6e7681; --tfm-layer-panels: #8957e5; --tfm-layer-nids_mpts: #3fb950; --tfm-layer-onts: #2ea043; --tfm-layer-field_issues: #f85149; --tfm-layer-field_markups: #e3b341; --tfm-layer-olt_devices: #db61a2; }