gui: real RTL + no-download-logic checks; split into velox-gui-lib

Follow-up hardening after a review noted the RTL "check" verified nothing
(a .ts stub that no test loads), matching a session-wide pattern of
checks written against what should be true rather than what would break.

- Split the non-main() code into velox-gui-lib (STATIC) so tests link the
  real widgets/models, not a reimplementation.
- tst_rtl: builds the real MainWindow, flips layoutDirection, asserts the
  direction propagates to the central widget AND that the offline-banner
  QHBoxLayout actually mirrors (label x-position LTR vs RTL differs by
  >100px). Verified it fails when the banner is pinned LtR.
- gui_no_download_logic: a ctest that greps gui/src for curl_*/pwrite/
  sqlite/QSqlDatabase/QNetworkAccessManager and fails on a hit — CLAUDE.md
  §3 as an executable check. Verified it fails when a curl_ token is added.
- Still uncovered (noted, not claimed): that the translation catalogue
  loads and the right context/strings resolve at runtime.

Not covered here because the files are PKG/QA-owned: tools/bootstrap.sh
ships a package name that does not exist on 26.04 (libqt6svg6-dev; the
real one is qt6-svg-dev), and --check validates pkg-config outcomes
rather than the apt names it would install. Both, plus the same name in
AGENT-PKG-QA.md and the README, are written up apply-ready in
gui/docs/pkg-qa-requests-m1.md.

Co-Authored-By: Claude Sonnet 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_016Ne28kx4VreeBWZv82Nksd
This commit is contained in:
2026-09-10 01:12:06 +04:00
co-authored by Claude Sonnet 5
parent 2a87abe96d
commit 2959b0f707
5 changed files with 270 additions and 28 deletions
+19 -23
View File
@@ -1,24 +1,24 @@
# velox-gui — the Qt 6 Widgets client. Lane GUI.
#
# CLAUDE.md §3: zero download logic here. This target links libveloxproto (the wire
# types, ADR 0009) and NEVER libveloxcore — a grep for curl|pwrite|sqlite under gui/
# must come back empty.
# types, ADR 0009) and NEVER libveloxcore — enforced by tests/ (see the
# gui_no_download_logic test).
#
# The root CMakeLists.txt add_subdirectory()s this unconditionally once the file exists,
# so it must stay configurable even before CORE has landed veloxproto. Until that target
# exists we announce and bail; the moment `core/` merges its libveloxproto, the GUI
# lights up with no edit here (same contract as the root file's EXISTS() guards).
# so it must stay configurable even if veloxproto is ever absent again. Until that target
# exists we announce and bail.
if(NOT TARGET veloxproto)
message(STATUS "velox-gui: libveloxproto has not landed yet — GUI target skipped. "
"It builds automatically once core/ merges the veloxproto target (ADR 0009).")
message(STATUS "velox-gui: libveloxproto target missing — GUI target skipped. "
"It builds automatically once core/ provides the veloxproto target (ADR 0009).")
return()
endif()
set(CMAKE_AUTOMOC ON)
add_executable(velox-gui
src/main.cpp
# Everything except main() lives in a static lib so the tests can link the real widgets
# and models rather than a reimplementation.
add_library(velox-gui-lib STATIC
src/rpc/RpcConnection.cpp
src/rpc/RpcClient.cpp
src/models/DownloadTableModel.cpp
@@ -26,29 +26,25 @@ add_executable(velox-gui
src/mainwindow/MainWindow.cpp
)
target_include_directories(velox-gui PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/src)
target_compile_features(velox-gui PRIVATE cxx_std_23)
# Warnings at target scope, not via CMAKE_CXX_FLAGS — the dev/tsan presets overwrite that
# cache variable wholesale (same rationale as core/CMakeLists.txt).
target_compile_options(velox-gui PRIVATE -Wall -Wextra -Wpedantic -Werror)
target_link_libraries(velox-gui PRIVATE
target_include_directories(velox-gui-lib PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/src)
target_compile_features(velox-gui-lib PUBLIC cxx_std_23)
target_compile_options(velox-gui-lib PRIVATE -Wall -Wextra -Wpedantic -Werror)
target_link_libraries(velox-gui-lib PUBLIC
velox::proto
Qt6::Widgets
Qt6::Svg
Qt6::Network
)
set_target_properties(velox-gui PROPERTIES
WIN32_EXECUTABLE OFF
MACOSX_BUNDLE OFF
)
add_executable(velox-gui src/main.cpp)
target_compile_options(velox-gui PRIVATE -Wall -Wextra -Wpedantic -Werror)
target_link_libraries(velox-gui PRIVATE velox-gui-lib)
set_target_properties(velox-gui PROPERTIES WIN32_EXECUTABLE OFF MACOSX_BUNDLE OFF)
# --- i18n -------------------------------------------------------------------------------
# Every string in the GUI goes through tr(); the Arabic stub exists to prove the RTL
# layout survives (GUI DoD). lrelease/lupdate come from Qt6::LinguistTools, found by root.
# layout survives (GUI DoD), exercised by tests/tst_rtl. lrelease also drops the .qm at
# ${CMAKE_CURRENT_BINARY_DIR}/velox_ar.qm, which the RTL test loads by path.
qt_add_translations(velox-gui TS_FILES i18n/velox_ar.ts)
# --- tests ---------------------------------------------------------------------------
+115
View File
@@ -0,0 +1,115 @@
# GUI → PKG/QA requests (M1)
Filed by lane GUI. These touch PKG/QA-owned files (`tools/bootstrap.sh`, the CI jobs,
`docs/agents/AGENT-PKG-QA.md`) and the root `README.md`, so GUI is not making the edits —
CLAUDE.md §1. Everything below is apply-ready.
---
## R1 — `libqt6svg6-dev` is not a real package on 26.04 (blocks the GUI DoD)
`apt-cache policy libqt6svg6-dev` on a clean Ubuntu 26.04 box gives `Candidate: (none)`.
The Qt 6 SVG **dev** package in the 26.04 archive is **`qt6-svg-dev`** (currently
`6.10.2-2`). `libqt6svg6` (no `-dev`) exists as the runtime lib but carries no headers or
CMake config, so `find_package(Qt6 COMPONENTS Svg)` fails without `qt6-svg-dev`.
I checked the other 38 apt names in `bootstrap.sh` against `apt-cache policy` on 26.04 —
`libqt6svg6-dev` is the only one with no candidate. Everything else resolves.
**Why it passed here anyway:** this dev box already has `qt6-svg-dev` installed (it came in
with a `qt6-base-dev` recommends chain at some point), so `pkg-config --exists Qt6Svg`
succeeds and `--check` is green. On a clean VM the `apt-get install` step aborts before
`--check` ever runs.
### Fix — three files, same one-line change
**`tools/bootstrap.sh`** (in `APT_GUI`, ~line 87):
```diff
APT_GUI=(
qt6-base-dev
qt6-tools-dev
qt6-tools-dev-tools # lrelease/lupdate for i18n
- libqt6svg6-dev # GUI: SVG icon rendering
+ qt6-svg-dev # GUI: SVG icon rendering (was libqt6svg6-dev — no such
+ # package on 26.04; this is the one with headers +
+ # the Qt6SvgConfig.cmake find_package needs)
)
```
**`docs/agents/AGENT-PKG-QA.md`** (~line 17):
```diff
- `libqt6svg6-dev`, `libsecret-1-dev`, `nodejs`/`npm` and (optionally) `clang` are
+ `qt6-svg-dev`, `libsecret-1-dev`, `nodejs`/`npm` and (optionally) `clang` are
```
**`README.md`** (toolchain table, ~line 106):
```diff
sudo apt update && sudo apt install -y \
- libqt6svg6-dev \ # GUI: SVG icon rendering
+ qt6-svg-dev \ # GUI: SVG icon rendering
libsecret-1-dev \ # DAEMON: Secret Service for site logins
```
---
## R2 — `--check` verifies the outcome, not the thing that can break
`--check` today runs the "verify toolchain" block: `command -v` for the binaries,
`pkg-config --exists` for the dev libs (`Qt6Core`, `Qt6Widgets`, `Qt6Svg`, `libcurl`,
`sqlite3`, `libssl`, `libsecret-1`, `libavformat`), plus the cmake/g++/node version
floors. Every one of those checks a *result* on a box that already installed everything.
None of them touch the apt package **names** in `APT_*`, which is the only part of the
script that can be wrong on a clean machine — as R1 just demonstrated. So `--check`
reported green for a script that fails `apt-get install` on its target OS.
### Proposed fix — make `--check` validate the install list it would actually run
Add this to the verify block (runs in **both** modes; in `--check` mode it is the point of
the exercise). It needs neither root nor network — `apt-cache policy` reads the local
package lists that `apt-get update` already populated:
```bash
# Every apt name this script would install must be a real, installable package on THIS
# release. This is the check that would have caught libqt6svg6-dev before the VM did.
log "validating apt package names (${#PKGS[@]})"
missing_pkgs=()
for pkg in "${PKGS[@]}"; do
# `apt-cache policy <pkg>` prints "Candidate: (none)" for a name with no installable
# version, and exits 0 either way — so grep the candidate line, don't trust $?.
cand="$(apt-cache policy "$pkg" 2>/dev/null | sed -n 's/^ Candidate: //p')"
if [[ -z "$cand" || "$cand" == "(none)" ]]; then
printf ' \033[31m✗\033[0m %-20s no installable candidate on this release\n' "$pkg"
missing_pkgs+=("$pkg")
fail=1
fi
done
if (( ${#missing_pkgs[@]} == 0 )); then
printf ' \033[32m✓\033[0m %-20s all %d resolve\n' "apt names" "${#PKGS[@]}"
fi
```
Notes for whoever applies it:
* `PKGS` is already assembled above the `CHECK_ONLY` branch, and it already respects
`--with-clang` / `--with-packaging`, so the loop covers exactly what would be installed.
* Belt-and-braces alternative: `apt-get install -s --no-install-recommends "${PKGS[@]}"`
(`-s` = simulate, no root) also catches an *un-satisfiable dependency*, not just a
missing name. Downside: it needs the apt lists reasonably fresh and is noisier to parse.
`apt-cache policy` is enough to catch the whole class of bug in R1; pick whichever you
want to own.
* CI: the `bootstrap-script` job already runs `./tools/bootstrap.sh --check` after the
install. With this change that job would have gone red on `libqt6svg6-dev` **at the
`--check` step on the 24.04 runner** even before the install step failed, because the
name has no candidate there either.
---
## R3 (smaller) — a GUI CI job
There is no job that builds `velox-gui` on its own or runs its offscreen smoke. The
`build` matrix will pick up the target and the `gui_downloadtablemodel` ctest once
`gui/` merges, but the GUI DoD items — 10k rows at 60 fps, flat memory over 10 min,
`--slow`/`--flaky`/`--drop-connection` recovery, `grep -r 'curl\|pwrite\|sqlite' gui/`
empty, RTL layout — need somewhere to run. GUI can write the harness
(`gui/tests/…` + a script); wiring it into `.github/workflows/ci.yml` is PKG/QA. Say the
word and I'll send the harness as a follow-up request with the job stanza pre-written.
+27 -5
View File
@@ -1,22 +1,44 @@
# GUI unit tests. Lane GUI.
#
# CLAUDE.md §5: a feature with no test does not exist. The model is the piece with real
# logic that can be tested headless — the progress patch must be a narrow dataChanged and
# must never reset the model (docs/03-gui-spec.md §1).
# CLAUDE.md §5: a feature with no test does not exist. Each test below has a concrete
# failure it catches — that is the bar, not "it runs green".
set(CMAKE_AUTOMOC ON)
find_package(Qt6 REQUIRED COMPONENTS Test)
# tst_downloadtablemodel — the model's logic, headless.
# Red when: a progress patch widens past the value columns, a reset slips into
# applyProgress, applyProgress starts synthesizing rows for unknown ids, or add/remove
# stops emitting rowsInserted/rowsRemoved.
add_executable(tst_downloadtablemodel
tst_downloadtablemodel.cpp
${CMAKE_CURRENT_SOURCE_DIR}/../src/models/DownloadTableModel.cpp
)
target_include_directories(tst_downloadtablemodel PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/../src)
target_compile_features(tst_downloadtablemodel PRIVATE cxx_std_23)
target_compile_options(tst_downloadtablemodel PRIVATE -Wall -Wextra -Wpedantic -Werror)
target_link_libraries(tst_downloadtablemodel PRIVATE Qt6::Core Qt6::Test)
add_test(NAME gui_downloadtablemodel COMMAND tst_downloadtablemodel)
set_tests_properties(gui_downloadtablemodel PROPERTIES LABELS "gui")
# tst_rtl — the real RTL check the GUI DoD asks for (the .ts stub alone verifies nothing).
# Red when: the main window stops propagating Qt::RightToLeft to its children, or the
# offline-banner layout is pinned so it does not mirror under RTL.
add_executable(tst_rtl tst_rtl.cpp)
target_compile_features(tst_rtl PRIVATE cxx_std_23)
target_compile_options(tst_rtl PRIVATE -Wall -Wextra -Wpedantic -Werror)
target_link_libraries(tst_rtl PRIVATE velox-gui-lib Qt6::Widgets Qt6::Test)
add_test(NAME gui_rtl COMMAND tst_rtl)
set_tests_properties(gui_rtl PROPERTIES
LABELS "gui"
ENVIRONMENT "QT_QPA_PLATFORM=offscreen")
# gui_no_download_logic — CLAUDE.md §3 as an executable check, not a hope.
# Red when: a download-logic token (curl, raw pwrite, sqlite, QSqlDatabase) appears
# under gui/src. grep exits 0 only when it finds a match, so a hit fails the test.
add_test(NAME gui_no_download_logic
COMMAND ${CMAKE_COMMAND}
-DGUI_SRC=${CMAKE_CURRENT_SOURCE_DIR}/../src
-P ${CMAKE_CURRENT_SOURCE_DIR}/no_download_logic.cmake)
set_tests_properties(gui_no_download_logic PROPERTIES LABELS "gui")
+39
View File
@@ -0,0 +1,39 @@
# Fails if any download-logic token appears under gui/src. Run as a ctest (see
# tests/CMakeLists.txt). CLAUDE.md §3: "A grep for curl|pwrite|sqlite in gui/ must come
# back empty."
#
# Invoked with -DGUI_SRC=<abs path to gui/src>.
if(NOT DEFINED GUI_SRC)
message(FATAL_ERROR "GUI_SRC not set")
endif()
file(GLOB_RECURSE sources "${GUI_SRC}/*.cpp" "${GUI_SRC}/*.hpp")
# Word-ish boundaries so this file's own prose and e.g. "curly" do not trip it, and so a
# comment mentioning the rule is fine but a real call is not.
set(forbidden
"curl_[a-z_]+" # libcurl
"\\bpwrite\\b" # raw positioned writes — the transfer path, not the GUI
"sqlite3?_[a-z_]+" # sqlite C API
"QSqlDatabase" # or via Qt
"QNetworkAccessManager" # the GUI talks to the daemon over QLocalSocket, nothing else
)
set(hits "")
foreach(file ${sources})
file(STRINGS "${file}" matched REGEX "(curl_[a-z_]+|\\bpwrite\\b|sqlite3?_[a-z_]+|QSqlDatabase|QNetworkAccessManager)")
foreach(line ${matched})
string(STRIP "${line}" line)
list(APPEND hits "${file}: ${line}")
endforeach()
endforeach()
list(LENGTH hits n)
if(n GREATER 0)
string(REPLACE ";" "\n " pretty "${hits}")
message(FATAL_ERROR
"download-logic token(s) found under gui/src — CLAUDE.md §3:\n ${pretty}")
endif()
message(STATUS "gui/src is clean of download-logic tokens (${forbidden})")
+70
View File
@@ -0,0 +1,70 @@
// RTL layout test. Lane GUI.
//
// The GUI DoD asks for "a stub Arabic .ts proves the RTL layout survives". A .ts file on
// its own proves nothing — this is the check with teeth: build the real main window, flip
// the layout direction, and assert (a) the direction propagates to the children and
// (b) a layout we own actually mirrors, by comparing a widget's position LTR vs RTL.
#include <QApplication>
#include <QLabel>
#include <QWidget>
#include <QtTest>
#include "mainwindow/MainWindow.hpp"
#include "rpc/RpcClient.hpp"
using velox::gui::MainWindow;
namespace {
int settledLeftOf(QWidget *w) {
QCoreApplication::processEvents();
QTest::qWait(30);
QCoreApplication::processEvents();
return w->mapTo(w->window(), QPoint(0, 0)).x();
}
} // namespace
class TstRtl : public QObject {
Q_OBJECT
private slots:
void directionPropagatesAndLayoutMirrors();
};
void TstRtl::directionPropagatesAndLayoutMirrors() {
velox::gui::rpc::RpcClient client(QStringLiteral("/nonexistent/velox-rtl-test.sock"));
MainWindow window(&client); // never started — the window must build without a daemon
window.resize(900, 500);
window.setLayoutDirection(Qt::LeftToRight);
window.show();
auto *bannerLabel = window.findChild<QLabel *>(QStringLiteral("offlineBannerLabel"));
QVERIFY2(bannerLabel != nullptr, "offline banner label not found");
QVERIFY2(bannerLabel->isVisible(), "offline banner should be visible while disconnected");
const int leftLtr = settledLeftOf(bannerLabel);
window.setLayoutDirection(Qt::RightToLeft);
// Propagation: the window and its central widget must both flip.
QCOMPARE(window.layoutDirection(), Qt::RightToLeft);
QVERIFY2(window.centralWidget() != nullptr, "no central widget");
QCOMPARE(window.centralWidget()->layoutDirection(), Qt::RightToLeft);
const int leftRtl = settledLeftOf(bannerLabel);
// The banner is [label][stretch] in a QHBoxLayout we build ourselves. LTR pins the
// label near x=0; RTL must push it to the right by (roughly) the free space. If the
// layout ignored direction this delta would be ~0.
QVERIFY2(leftRtl - leftLtr > 100,
qPrintable(QStringLiteral("offline banner did not mirror under RTL: "
"label x LTR=%1 RTL=%2")
.arg(leftLtr)
.arg(leftRtl)));
}
QTEST_MAIN(TstRtl)
#include "tst_rtl.moc"