aboutsummaryrefslogtreecommitdiffstats
path: root/src/lib/corelib/loader/productsresolver.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'src/lib/corelib/loader/productsresolver.cpp')
-rw-r--r--src/lib/corelib/loader/productsresolver.cpp599
1 files changed, 599 insertions, 0 deletions
diff --git a/src/lib/corelib/loader/productsresolver.cpp b/src/lib/corelib/loader/productsresolver.cpp
new file mode 100644
index 000000000..997a4dc57
--- /dev/null
+++ b/src/lib/corelib/loader/productsresolver.cpp
@@ -0,0 +1,599 @@
+/****************************************************************************
+**
+** Copyright (C) 2023 The Qt Company Ltd.
+** Contact: https://www.qt.io/licensing/
+**
+** This file is part of Qbs.
+**
+** $QT_BEGIN_LICENSE:LGPL$
+** Commercial License Usage
+** Licensees holding valid commercial Qt licenses may use this file in
+** accordance with the commercial license agreement provided with the
+** Software or, alternatively, in accordance with the terms contained in
+** a written agreement between you and The Qt Company. For licensing terms
+** and conditions see https://www.qt.io/terms-conditions. For further
+** information use the contact form at https://www.qt.io/contact-us.
+**
+** GNU Lesser General Public License Usage
+** Alternatively, this file may be used under the terms of the GNU Lesser
+** General Public License version 3 as published by the Free Software
+** Foundation and appearing in the file LICENSE.LGPL3 included in the
+** packaging of this file. Please review the following information to
+** ensure the GNU Lesser General Public License version 3 requirements
+** will be met: https://www.gnu.org/licenses/lgpl-3.0.html.
+**
+** GNU General Public License Usage
+** Alternatively, this file may be used under the terms of the GNU
+** General Public License version 2.0 or (at your option) the GNU General
+** Public license version 3 or any later version approved by the KDE Free
+** Qt Foundation. The licenses are as published by the Free Software
+** Foundation and appearing in the file LICENSE.GPL2 and LICENSE.GPL3
+** included in the packaging of this file. Please review the following
+** information to ensure the GNU General Public License requirements will
+** be met: https://www.gnu.org/licenses/gpl-2.0.html and
+** https://www.gnu.org/licenses/gpl-3.0.html.
+**
+** $QT_END_LICENSE$
+**
+****************************************************************************/
+
+#include "productsresolver.h"
+
+#include "itemreader.h"
+#include "loaderutils.h"
+#include "productresolver.h"
+
+#include <language/language.h>
+#include <language/scriptengine.h>
+#include <logging/categories.h>
+#include <logging/translator.h>
+#include <tools/profiling.h>
+#include <tools/progressobserver.h>
+#include <tools/setupprojectparameters.h>
+#include <tools/stringconstants.h>
+
+#include <algorithm>
+#include <condition_variable>
+#include <future>
+#include <mutex>
+#include <queue>
+#include <system_error>
+#include <thread>
+#include <unordered_map>
+#include <vector>
+
+namespace qbs::Internal {
+namespace {
+struct ThreadInfo {
+ ThreadInfo(std::future<void> &&future, LoaderState &loaderState)
+ : future(std::move(future)), loaderState(loaderState)
+ {}
+ std::future<void> future;
+ LoaderState &loaderState;
+ bool done = false;
+};
+
+struct ProductWithLoaderState {
+ ProductWithLoaderState(ProductContext &product, LoaderState *loaderState)
+ : product(&product), loaderState(loaderState) {}
+ ProductContext * const product;
+ LoaderState *loaderState;
+};
+
+class ThreadsLocker {
+public:
+ ThreadsLocker(std::launch mode, std::mutex &mutex) {
+ if (mode == std::launch::async)
+ lock = std::make_unique<std::unique_lock<std::mutex>>(mutex);
+ }
+ std::unique_ptr<std::unique_lock<std::mutex>> lock;
+};
+} // namespace
+
+class ProductsResolver
+{
+public:
+ ProductsResolver(LoaderState &loaderState) : m_loaderState(loaderState) {}
+ void resolve();
+
+private:
+ void initialize();
+ void initializeProductQueue();
+ void initializeLoaderStatePool();
+ void runScheduler();
+ void scheduleNext();
+ bool tryToReserveLoaderState(ProductWithLoaderState &product, Deferral deferral);
+ std::optional<std::pair<ProductContext *, Deferral>>
+ unblockProductWaitingForLoaderState(LoaderState &loaderState);
+ void startJob(const ProductWithLoaderState &product, Deferral deferral);
+ void checkForCancelation();
+ void handleFinishedThreads();
+ void queueProductForScheduling(const ProductWithLoaderState &product, Deferral deferral);
+ void waitForSingleDependency(const ProductWithLoaderState &product, ProductContext &dependency);
+ void waitForBulkDependency(const ProductWithLoaderState &product);
+ void unblockProductsWaitingForDependency(ProductContext &finishedProduct);
+ void postProcess();
+ void checkForMissedBulkDependencies(const ProductContext &product);
+
+ static int dependsItemCount(const Item *item);
+ static int dependsItemCount(ProductContext &product);
+
+ LoaderState &m_loaderState;
+ std::queue<std::pair<ProductWithLoaderState, int>> m_productsToSchedule;
+ std::vector<ProductContext *> m_finishedProducts;
+ std::unordered_map<ProductContext *,
+ std::vector<ProductWithLoaderState>> m_waitingForSingleDependency;
+ std::vector<ProductWithLoaderState> m_waitingForBulkDependency;
+ std::unordered_map<LoaderState *, std::queue<std::pair<ProductContext *, Deferral>>> m_waitingForLoaderState;
+ std::unordered_map<ProductContext *, ThreadInfo> m_runningThreads;
+ std::mutex m_threadsMutex;
+ std::condition_variable m_threadsNotifier;
+ std::vector<std::unique_ptr<ScriptEngine>> m_enginePool;
+ std::vector<std::unique_ptr<LoaderState>> m_loaderStatePool;
+ std::vector<LoaderState *> m_availableLoaderStates;
+ std::mutex m_cancelingMutex;
+ std::launch m_asyncMode = std::launch::async;
+ int m_maxJobCount = m_loaderState.parameters().maxJobCount();
+ bool m_canceling = false;
+};
+
+void resolveProducts(LoaderState &loaderState)
+{
+ ProductsResolver(loaderState).resolve();
+}
+
+void ProductsResolver::resolve()
+{
+ initialize();
+ try {
+ runScheduler();
+ } catch (const ErrorInfo &e) {
+ for (auto &thread : m_runningThreads)
+ thread.second.future.wait();
+ throw e;
+ }
+ postProcess();
+}
+
+void ProductsResolver::initialize()
+{
+ initializeProductQueue();
+ initializeLoaderStatePool();
+}
+
+void ProductsResolver::initializeProductQueue()
+{
+ TopLevelProjectContext &topLevelProject = m_loaderState.topLevelProject();
+ std::vector<ProductContext *> sortedProducts;
+ for (ProjectContext * const projectContext : topLevelProject.projects()) {
+ for (ProductContext &product : projectContext->products) {
+ topLevelProject.addProductToHandle(product);
+ const auto it = std::lower_bound(sortedProducts.begin(), sortedProducts.end(), product,
+ [&product](ProductContext *p1, const ProductContext &) {
+ return dependsItemCount(*p1) < dependsItemCount(product);
+ });
+ sortedProducts.insert(it, &product);
+ }
+ }
+
+ for (ProductContext * const product : sortedProducts) {
+ queueProductForScheduling(ProductWithLoaderState(*product, nullptr), Deferral::Allowed);
+ if (product->shadowProduct) {
+ topLevelProject.addProductToHandle(*product->shadowProduct);
+ queueProductForScheduling(ProductWithLoaderState(*product->shadowProduct, nullptr),
+ Deferral::Allowed);
+ }
+ }
+}
+
+void ProductsResolver::initializeLoaderStatePool()
+{
+ TopLevelProjectContext &topLevelProject = m_loaderState.topLevelProject();
+
+ // Adapt max job count: It makes no sense to have it be higher than the number of products
+ // or what can actually be run concurrently. In both cases, we would simply waste resources.
+ const int maxConcurrency = std::thread::hardware_concurrency();
+ if (maxConcurrency > 0 && maxConcurrency < m_maxJobCount)
+ m_maxJobCount = maxConcurrency;
+ if (m_maxJobCount > topLevelProject.productsToHandleCount())
+ m_maxJobCount = topLevelProject.productsToHandleCount();
+
+ // The number of engines and loader states we need to allocate here is one less than the
+ // total number of concurrent jobs, as we already have one loader state that we can re-use.
+ if (m_maxJobCount > 1)
+ m_enginePool.reserve(m_maxJobCount - 1);
+ m_loaderStatePool.reserve(m_enginePool.size());
+ m_availableLoaderStates.reserve(m_enginePool.size() + 1);
+ m_availableLoaderStates.push_back(&m_loaderState);
+ for (std::size_t i = 0; i < m_enginePool.capacity(); ++i) {
+ ScriptEngine &engine = *m_enginePool.emplace_back(
+ ScriptEngine::create(m_loaderState.logger(), EvalContext::PropertyEvaluation));
+ ItemPool &itemPool = topLevelProject.createItemPool();
+ engine.setEnvironment(m_loaderState.parameters().adjustedEnvironment());
+ auto loaderState = std::make_unique<LoaderState>(
+ m_loaderState.parameters(), topLevelProject, itemPool, engine,
+ m_loaderState.logger());
+ m_loaderStatePool.push_back(std::move(loaderState));
+ m_availableLoaderStates.push_back(m_loaderStatePool.back().get());
+ if (topLevelProject.progressObserver())
+ topLevelProject.progressObserver()->addScriptEngine(m_enginePool.back().get());
+ }
+ qCDebug(lcLoaderScheduling) << "using" << m_availableLoaderStates.size() << "loader states";
+ if (int(m_availableLoaderStates.size()) == 1)
+ m_asyncMode = std::launch::deferred;
+}
+
+void ProductsResolver::runScheduler()
+{
+ AccumulatingTimer timer(m_loaderState.parameters().logElapsedTime()
+ ? &m_loaderState.topLevelProject().timingData().resolvingProducts
+ : nullptr);
+
+ ThreadsLocker threadsLock(m_asyncMode, m_threadsMutex);
+ while (true) {
+ scheduleNext();
+ if (m_runningThreads.empty())
+ break;
+ if (m_asyncMode == std::launch::async) {
+ qCDebug(lcLoaderScheduling()) << "scheduling paused, waiting for threads to finish";
+ m_threadsNotifier.wait(*threadsLock.lock);
+ }
+ checkForCancelation();
+ handleFinishedThreads();
+ }
+
+ QBS_CHECK(m_productsToSchedule.empty());
+ QBS_CHECK(m_loaderState.topLevelProject().productsToHandleCount() == 0);
+ QBS_CHECK(m_runningThreads.empty());
+ QBS_CHECK(m_waitingForSingleDependency.empty());
+ QBS_CHECK(m_waitingForBulkDependency.empty());
+}
+
+void ProductsResolver::scheduleNext()
+{
+ TopLevelProjectContext &topLevelProject = m_loaderState.topLevelProject();
+ AccumulatingTimer timer(m_loaderState.parameters().logElapsedTime()
+ ? &topLevelProject.timingData().schedulingProducts : nullptr);
+ while (m_maxJobCount > int(m_runningThreads.size()) && !m_productsToSchedule.empty()) {
+ auto [product, toHandleCountOnInsert] = m_productsToSchedule.front();
+ m_productsToSchedule.pop();
+
+ qCDebug(lcLoaderScheduling) << "potentially scheduling product"
+ << product.product->displayName()
+ << "unhandled product count on queue insertion"
+ << toHandleCountOnInsert << "current unhandled product count"
+ << topLevelProject.productsToHandleCount();
+
+ // If the number of unfinished products has shrunk since the last time we tried handling
+ // this product, there has been forward progress and we can allow a deferral.
+ const Deferral deferral = toHandleCountOnInsert == -1
+ || toHandleCountOnInsert > topLevelProject.productsToHandleCount()
+ ? Deferral::Allowed : Deferral::NotAllowed;
+
+ if (!tryToReserveLoaderState(product, deferral))
+ continue;
+
+ startJob(product, deferral);
+ }
+
+ // There are jobs running, so forward progress is still possible.
+ if (!m_runningThreads.empty())
+ return;
+
+ QBS_CHECK(m_productsToSchedule.empty());
+
+ // If we end up here, nothing was scheduled in the loop above, which means that either ...
+ // a) ... we are done or
+ // b) ... we finally need to schedule our bulk dependencies or
+ // c) ... we need to schedule products waiting for an unhandled dependency.
+ // In the latter case, the project has at least one dependency cycle, and the
+ // DependencyResolver will emit an error.
+
+ // a)
+ if (m_waitingForBulkDependency.empty() && m_waitingForSingleDependency.empty())
+ return;
+
+ // b)
+ for (const ProductWithLoaderState &product : m_waitingForBulkDependency)
+ queueProductForScheduling(product, Deferral::NotAllowed);
+ if (!m_productsToSchedule.empty()) {
+ m_waitingForBulkDependency.clear();
+ scheduleNext();
+ return;
+ }
+
+ // c)
+ for (const auto &e : m_waitingForSingleDependency) {
+ for (const ProductWithLoaderState &p : e.second)
+ queueProductForScheduling(p, Deferral::NotAllowed);
+ }
+ QBS_CHECK(!m_productsToSchedule.empty());
+ scheduleNext();
+}
+
+bool ProductsResolver::tryToReserveLoaderState(ProductWithLoaderState &product, Deferral deferral)
+{
+ QBS_CHECK(!m_availableLoaderStates.empty());
+ if (!product.loaderState) {
+ product.loaderState = m_availableLoaderStates.back();
+ m_availableLoaderStates.pop_back();
+ return true;
+ }
+ if (const auto it = std::find(m_availableLoaderStates.begin(), m_availableLoaderStates.end(),
+ product.loaderState); it != m_availableLoaderStates.end()) {
+ m_availableLoaderStates.erase(it);
+ return true;
+ }
+ qCDebug(lcLoaderScheduling) << "loader state" << product.loaderState << " for product"
+ << product.product->displayName()
+ << "not available, adding product to wait queue";
+ m_waitingForLoaderState[product.loaderState].push({product.product, deferral});
+ return false;
+}
+
+std::optional<std::pair<ProductContext *, Deferral>>
+ProductsResolver::unblockProductWaitingForLoaderState(LoaderState &loaderState)
+{
+ auto &waitingForLoaderState = m_waitingForLoaderState[&loaderState];
+ if (waitingForLoaderState.empty())
+ return {};
+ const auto product = waitingForLoaderState.front();
+ waitingForLoaderState.pop();
+ qCDebug(lcLoaderScheduling) << "loader state" << &loaderState << "now available for product"
+ << product.first->displayName();
+ return product;
+}
+
+void ProductsResolver::startJob(const ProductWithLoaderState &product, Deferral deferral)
+{
+ QBS_CHECK(product.loaderState);
+ qCDebug(lcLoaderScheduling) << "scheduling product" << product.product->displayName()
+ << "with loader state" << product.loaderState
+ << "and deferral mode" << int(deferral);
+ try {
+ const auto it = m_runningThreads.emplace(product.product, ThreadInfo(std::async(m_asyncMode,
+ [this, product, deferral] {
+ product.loaderState->itemReader().setExtraSearchPathsStack(
+ product.product->project->searchPathsStack);
+ resolveProduct(*product.product, deferral, *product.loaderState);
+
+ // The search paths stack can change during dependency resolution
+ // (due to module providers); check that we've rolled back all the changes
+ QBS_CHECK(product.loaderState->itemReader().extraSearchPathsStack()
+ == product.product->project->searchPathsStack);
+
+ std::lock_guard cancelingLock(m_cancelingMutex);
+ if (m_canceling)
+ return;
+ ThreadsLocker threadsLock(m_asyncMode, m_threadsMutex);
+ if (const auto it = m_runningThreads.find(product.product);
+ it != m_runningThreads.end()) {
+ it->second.done = true;
+ qCDebug(lcLoaderScheduling) << "thread for product"
+ << product.product->displayName()
+ << "finished, waking up scheduler";
+ m_threadsNotifier.notify_one();
+ }
+ }), *product.loaderState));
+
+ // With just one worker thread, the notify/wait overhead would be excessive, so
+ // we run the task synchronously.
+ if (m_asyncMode == std::launch::deferred) {
+ qCDebug(lcLoaderScheduling) << "blocking on product thread immediately";
+ it.first->second.future.wait();
+ }
+ } catch (const std::system_error &e) {
+ if (e.code() != std::errc::resource_unavailable_try_again)
+ throw e;
+ qCWarning(lcLoaderScheduling) << "failed to create thread";
+ if (m_maxJobCount >= 2) {
+ m_maxJobCount /= 2;
+ qCWarning(lcLoaderScheduling) << "throttling down to" << m_maxJobCount << "jobs";
+ }
+ queueProductForScheduling(product, deferral);
+ m_availableLoaderStates.push_back(product.loaderState);
+ }
+}
+
+void ProductsResolver::checkForCancelation()
+{
+ if (m_loaderState.topLevelProject().isCanceled()) {
+ m_cancelingMutex.lock();
+ m_canceling = true;
+ m_cancelingMutex.unlock();
+ for (auto &thread : m_runningThreads)
+ thread.second.future.wait();
+ throw CancelException();
+ }
+}
+
+void ProductsResolver::handleFinishedThreads()
+{
+ TopLevelProjectContext &topLevelProject = m_loaderState.topLevelProject();
+ AccumulatingTimer timer(m_loaderState.parameters().logElapsedTime()
+ ? &topLevelProject.timingData().schedulingProducts : nullptr);
+
+ std::vector<std::pair<ProductWithLoaderState, Deferral>> productsToScheduleDirectly;
+ for (auto it = m_runningThreads.begin(); it != m_runningThreads.end();) {
+ ThreadInfo &ti = it->second;
+ if (!ti.done) {
+ ++it;
+ continue;
+ }
+ ti.future.wait();
+ ProductContext &product = *it->first;
+ LoaderState &loaderState = ti.loaderState;
+ it = m_runningThreads.erase(it);
+
+ qCDebug(lcLoaderScheduling) << "handling finished thread for product"
+ << product.displayName()
+ << "current unhandled product count is"
+ << topLevelProject.productsToHandleCount();
+
+ // If there are products waiting for the loader state used in the finished thread,
+ // we can start a job for one of them right away (but not in the loop,
+ // because startJob() modifies the thread list we are currently iterating over).
+ if (const auto productInfo = unblockProductWaitingForLoaderState(loaderState)) {
+ productsToScheduleDirectly.emplace_back(
+ ProductWithLoaderState(*productInfo->first, &loaderState),
+ productInfo->second);
+ } else {
+ qCDebug(lcLoaderScheduling) << "making loader state" << &loaderState
+ << "available again";
+ m_availableLoaderStates.push_back(&loaderState);
+ }
+
+ // If we encountered a dependency to an in-progress product or to a bulk dependency,
+ // we defer handling this product.
+ if (product.dependenciesResolvingPending()) {
+ qCDebug(lcLoaderScheduling) << "dependencies resolving not finished for product"
+ << product.displayName();
+ const auto pending = product.pendingDependency();
+ switch (pending.first) {
+ case ProductDependency::Single:
+ waitForSingleDependency(ProductWithLoaderState(product, &loaderState),
+ *pending.second);
+ break;
+ case ProductDependency::Bulk:
+ waitForBulkDependency(ProductWithLoaderState(product, &loaderState));
+ break;
+ case ProductDependency::None:
+ // This can happen if the dependency has finished in between the check in
+ // DependencyResolver and the one here.
+ QBS_CHECK(pending.second);
+ queueProductForScheduling(ProductWithLoaderState(product, &loaderState),
+ Deferral::Allowed);
+ break;
+ }
+ topLevelProject.incProductDeferrals();
+ } else {
+ qCDebug(lcLoaderScheduling) << "product" << product.displayName() << "finished";
+ topLevelProject.removeProductToHandle(product);
+ if (!product.name.startsWith(StringConstants::shadowProductPrefix()))
+ m_finishedProducts.push_back(&product);
+ topLevelProject.timingData() += product.timingData;
+ checkForMissedBulkDependencies(product);
+ topLevelProject.registerBulkDependencies(product);
+ unblockProductsWaitingForDependency(product);
+ }
+ }
+
+ for (const auto &productInfo : productsToScheduleDirectly)
+ startJob(productInfo.first, productInfo.second);
+}
+
+void ProductsResolver::queueProductForScheduling(const ProductWithLoaderState &product,
+ Deferral deferral)
+{
+ qCDebug(lcLoaderScheduling) << "queueing product" << product.product->displayName()
+ << "with deferral mode" << int(deferral);
+ m_productsToSchedule.emplace(product, deferral == Deferral::Allowed
+ ? -1 : m_loaderState.topLevelProject().productsToHandleCount());
+}
+
+void ProductsResolver::waitForSingleDependency(const ProductWithLoaderState &product,
+ ProductContext &dependency)
+{
+ qCDebug(lcLoaderScheduling) << "product" << product.product->displayName()
+ << "now waiting for single dependency"
+ << dependency.displayName();
+ m_waitingForSingleDependency[&dependency].push_back(product);
+}
+
+void ProductsResolver::waitForBulkDependency(const ProductWithLoaderState &product)
+{
+ qCDebug(lcLoaderScheduling) << "product" << product.product->displayName()
+ << "now waiting for bulk dependency";
+ m_waitingForBulkDependency.push_back(product);
+}
+
+void ProductsResolver::unblockProductsWaitingForDependency(ProductContext &finishedProduct)
+{
+ const auto it = m_waitingForSingleDependency.find(&finishedProduct);
+ if (it == m_waitingForSingleDependency.end())
+ return;
+
+ qCDebug(lcLoaderScheduling) << "unblocking all products waiting for now-finished product" <<
+ finishedProduct.displayName();
+ for (const ProductWithLoaderState &p : it->second) {
+ qCDebug(lcLoaderScheduling) << " unblocking product" << p.product->displayName();
+ queueProductForScheduling(p, Deferral::Allowed);
+ }
+ m_waitingForSingleDependency.erase(it);
+}
+
+void ProductsResolver::postProcess()
+{
+ for (ProductContext * const product : m_finishedProducts) {
+ if (product->product)
+ product->product->project->products.push_back(product->product);
+
+ // This has to be done in post-processing, because we need both product and shadow product
+ // to be ready, and contrary to what one might assume, there is no proper ordering
+ // between them regarding dependency resolving.
+ setupExports(*product, m_loaderState);
+ }
+
+ for (const auto &engine : m_enginePool)
+ m_loaderState.topLevelProject().collectDataFromEngine(*engine);
+
+ QBS_CHECK(!m_loaderState.topLevelProject().projects().empty());
+ const auto project = std::dynamic_pointer_cast<TopLevelProject>(
+ m_loaderState.topLevelProject().projects().front()->project);
+ QBS_CHECK(project);
+ for (LoaderState * const loaderState : m_availableLoaderStates)
+ project->warningsEncountered << loaderState->logger().warnings();
+}
+
+void ProductsResolver::checkForMissedBulkDependencies(const ProductContext &product)
+{
+ if (!product.product || !product.product->enabled || !product.bulkDependencies.empty())
+ return;
+ for (const FileTag &tag : product.product->fileTags) {
+ for (const auto &[p, location]
+ : m_loaderState.topLevelProject().finishedProductsWithBulkDependency(tag)) {
+ if (!p->product->enabled)
+ continue;
+ if (p->name == product.name)
+ continue;
+ ErrorInfo e;
+ e.append(Tr::tr("Cyclic dependencies detected:"));
+ e.append(Tr::tr("First product is '%1'.")
+ .arg(product.displayName()), product.item->location());
+ e.append(Tr::tr("Second product is '%1'.")
+ .arg(p->displayName()), p->item->location());
+ e.append(Tr::tr("Dependency from %1 to %2 was established via Depends.productTypes.")
+ .arg(p->displayName(), product.displayName()), location);
+ if (m_loaderState.parameters().productErrorMode() == ErrorHandlingMode::Strict)
+ throw e;
+ m_loaderState.logger().printWarning(e);
+ m_loaderState.logger().printWarning(
+ ErrorInfo(Tr::tr("Product '%1' had errors and was disabled.")
+ .arg(product.displayName()), product.item->location()));
+ m_loaderState.logger().printWarning(
+ ErrorInfo(Tr::tr("Product '%1' had errors and was disabled.")
+ .arg(p->displayName()), p->item->location()));
+ product.product->enabled = false;
+ p->product->enabled = false;
+ }
+ }
+}
+
+int ProductsResolver::dependsItemCount(const Item *item)
+{
+ int count = 0;
+ for (const Item * const child : item->children()) {
+ if (child->type() == ItemType::Depends)
+ ++count;
+ }
+ return count;
+}
+
+int ProductsResolver::dependsItemCount(ProductContext &product)
+{
+ if (product.dependsItemCount == -1)
+ product.dependsItemCount = dependsItemCount(product.item);
+ return product.dependsItemCount;
+}
+
+} // namespace qbs::Internal