Meta tags:
description= CLI tool for Angular. Contribute to angular/angular-cli development by creating an account on GitHub.;
Headings (most frequently used words):
uh, oh, comment, this, dherges, choose, reason, for, hiding, oct, 2026, code, test, angular, build, library, builder, left, navigation, pr, the, review, gemini, assist, bot, add, scaling, benchmark, suite, to, cli, ng, what, is, behavior, footer, commented, loading, 34290, menu, repository, 34290dherges, wants, merge, 15, commits, intoangular, mainangular, mainfrom, benchmarksdherges, benchmarkscopy, head, branch, name, clipboard, conversation, checklist, type, current, new, does, introduce, breaking, change, other, information, edited, there, was, an, error, while, please, reload, page, observations, along, way, caveats, 10, 12, 17, 23, 26, 27, reviewers, assignees, labels, projects, milestone, development, participant,
Text of the page (most frequently used words):
library (120), the (102), this (85), #builder (83), comment (68), build (62), inline (62), entry (61), flat (60), #angular (54), benchmark (48), packagr (46), was (44), and (44), there (43), run (43), while (38), for (38), done (37), reload (36), please (34), loading (34), error (33), page (33), scripts (31), benchmarks (30), resolved (29), reason (28), dherges (28), mts (27), test (27), 2026 (26), copy (24), thread (24), points (23), code (22), scaling (22), suite (22), add (21), changes (20), with (20), 500 (20), 1000 (19), all (18), choose (18), hiding (18), median (18), 200 (18), more (17), oct (17), 100 (17), that (15), style (15), point (15), time (14), fixup (14), github (13), per (13), review (12), learn (12), quality (12), index (12), reviewed (12), applied (11), pull (11), outdated (11), gemini (11), bot (11), reactions (11), sorry (11), something (11), went (11), wrong (11), markdown (11), link (11), external (11), cli (11), request (10), only (10), assist (10), csv (10), from (9), than (9), hide (9), low (9), duplicate (9), off (9), topic (9), abuse (9), spam (9), will (9), displayed (9), describe (9), others (9), problem (9), view (9), axis (9), count (9), flavor (9), new (9), entries (9), not (8), cannot (8), commented (8), native (8), fixtures (8), results (8), suggestion (7), suggestions (7), author (7), scss (7), layouts (7), layout (7), existing (6), one (6), across (6), roughly (6), scales (6), sweep (6), deep (6), css (6), october (6), iterations (6), sizes (6), you (5), can (5), security (5), left (5), sign (5), times (5), builds (5), against (5), support (5), directory (5), depth (5), far (5), better (5), into (5), cost (5), are (5), added (5), pnpm (5), 2000 (5), main (5), status (4), merge (4), line (4), marked (4), lines (4), have (4), file (4), what (4), open (4), options (4), number (4), full (4), 58s (4), small (4), readme (4), ran (4), 33s (4), 98s (4), 54s (4), 20s (4), 47s (4), 269 (4), branch (4), shape (4), behavior (4), commits (4), 2048 (4), documentation (4), 34290 (4), your (4), copilot (4), navigation (3), been (3), change (3), batch (3), single (3), commit (3), may (3), issues (3), conversation (3), hidden (3), characters (3), unicode (3), about (3), improvements (3), template (3), driver (3), templates (3), reaction (3), reacted (3), emoji (3), introduces (3), subsystem (3), measure (3), compare (3), various (3), feedback (3), out (3), devkit (3), admin (3), 10s (3), 93s (3), 21s (3), 16s (3), 03s (3), 122 (3), 01s (3), speedup (3), but (3), pattern (3), just (3), once (3), package (3), json (3), much (3), fixture (3), see (3), above (3), grows (3), costs (3), plus (3), adds (3), own (3), any (3), matrix (3), reported (3), design (3), its (3), climb (3), other (3), size (3), tested (3), cold (3), source (3), another (3), tab (3), window (3), refresh (3), session (3), developers (3), changelog (3), customer (3), enterprise (3), grade (3), features (3), information (2), docs (2), community (2), footer (2), now (2), check (2), back (2), comments (2), has (2), create (2), merging (2), these (2), development (2), milestone (2), yet (2), projects (2), area (2), reviewers (2), already (2), bidirectional (2), directories (2), path (2), dirname (2), flavour (2), statistical (2), variants (2), same (2), earlier (2), should (2), total (2), html (2), indicating (2), harness (2), performance (2), integration (2), highlights (2), several (2), robustness (2), parsing (2), pre (2), replacing (2), avoid (2), resource (2), ready (2), output (2), node (2), tmp (2), 1792749996 (2), 333 (2), 510 (2), 28ms (2), 57s (2), 283 (2), 67s (2), 252 (2), 78s (2), 28s (2), 240 (2), 75s (2), 245 (2), 70s (2), iteration (2), indicate (2), think (2), shared (2), thinking (2), investigated (2), cause (2), glob (2), matching (2), then (2), why (2), here (2), linear (2), higher (2), fixed (2), first (2), comparison (2), falling (2), additional (2), conclusion (2), keeps (2), ead4fa9 (2), force (2), pushed (2), recorded (2), implemented (2), node_path (2), each (2), resolve (2), require (2), deeply (2), nested (2), independent (2), curve (2), default (2), targets (2), zero (2), failures (2), track (2), closely (2), comparable (2), scale (2), structure (2), doesn (2), meaningful (2), unlike (2), past (2), decreases (2), monotonically (2), flattens (2), around (2), 6ms (2), regression (2), further (2), 46s (2), largest (2), 25s (2), 31s (2), 300 (2), symlinks (2), they (2), eyes (2), apf (2), under (2), execution (2), errors (2), i18n (2), packages (2), does (2), numbers (2), specific (2), reproducible (2), found (2), way (2), content (2), related (2), tests (2), message (2), wants (2), head (2), name (2), clipboard (2), insights (2), actions (2), 196 (2), repository (2), signed (2), universe (2), training (2), integrations (2), stories (2), center (2), business (2), fixes (2), products (2), secret (2), perform, action, share, personal, manage, cookies, contact, privacy, terms, inc, right, later, queued, multi, pending, reviews, must, order, valid, applying, deleted, supported, viewing, subset, closed, invalid, because, were, made, participant, successfully, close, none, labels, assigned, assignees, whose, approvals, affect, requirements, reviewer, account, join, free, show, contains, text, interpreted, compiled, differently, appears, below, editor, reveals, intermediate, missing, mkdirsync, recursive, true, writefilesync, formatcsvreport, measurements, adressed, geminie, 8166f3a, distribution, seems, noise, confirming, very, well, large, drawing, conclusions, read, varies, between, 65s, primary, stays, regardless, stylesheet, language, stay, consistent, beats, starting, counts, benchmarla, cross, 77f88c6, thumbs, legacy, includes, flavors, opportunities, improve, safety, runner, specifically, using, robust, regex, release, versions, finite, limit, prevent, memory, crashes, enforcing, upper, bounds, parameters, exhaustion, maxbuffer, infinity, reference, log, warnings, experimentalwarning, experimental, transform, types, 37s, 26s, 20ms, 14s, 59s, 50s, 79ms, 84s, 95ms, 72s, 87s, 15ms, 42s, 33ms, 02s, 81ms, 04s, 91s, 06ms, 120, 125, 11s, 16ms, 06s, 01ms, 274, 23s, 255, 79s, 98ms, written, 255787, 274234, 269976, 10005, 10059, 9700, 122579, 125114, 120750, 5913, 6038, 6029, 48019, 51275, 48161, 4466, 4783, 4420, 25214, 26671, 24539, 3869, 4721, 3928, 14197, 14574, 13840, 3500, 3539, 3592, 4979, 5102, 5142, 3262, 3365, 3332, durationms, entrypoints, styleflavor, details, today, knowledhe, attributed, different, architectures, sequential, approach, uses, compilation, host, worker, pools, afaik, most, likely, possible, bottleneck, don, actual, front, little, script, discovered, less, explain, gap, spent, some, achieves, throughput, subtracting, upfront, records, given, set, 295ms, 27x, 225ms, 248ms, 8ms, 270, 276, 260, report, bnchmark, 45eb7d1, d4d7b63, generated, either, project, benchmarking, fix, preparebuilderenv, creates, idempotently, tiny, symlink, mapping, core, architect, their, dist, combined, pointing, shim, dir, node_modules, where, ajv, formats, sass, esbuild, etc, correctly, linked, lets, spawned, every, bare, however, buildonce, takes, applies, nodepath, rather, clobbering, f7e154a, defaults, prior, since, multiplies, opt, list, d48bd74, 4a9b4ec, reuslts, baseline, libray, gracefully, detected, 8f7573d, 1d7a102, c7a812b, fe70c22, 90a5c08, 0aaa731, f760cb4, e8638b9, updated, covering, variations, 211, 209, dd25d8f, fdb8f5a, succeeded, appear, top, end, iterated, completed, 63s, possibly, tree, walk, overhead, showed, slightly, variance, itself, isn, raw, suggesting, shaped, both, 1024, 512, 256, 128, 30s, 103, 27s, 204, 29s, 18s, 400, 51s, 90s, 164, 324, 161, 124, draft, implementation, focuses, improving, dependency, resolution, symlinking, manual, store, updating, preventing, potential, crash, when, numeric, argument, passed, parseintlist, createrequire, questions, towards, references, including, generation, important, resolving, compatibility, caused, robustly, handling, broken, validating, inputs, logging, process, spawn, wrapping, block, ensure, temporary, always, cleaned, try, finally, depths, eexist, import, meta, label, robot, tooling, wiring, dispatcher, alongside, yes, introduce, breaking, absolute, machine, collected, portable, finding, hardware, intentionally, needs, runs, comparisons, shot, exercise, incremental, rebuilds, watch, mode, caveats, future, characteristic, independently, anyone, quoted, description, initital, presented, intersting, observation, dig, depper, result, created, running, shapes, through, even, such, growth, built, investigating, investigation, rises, exceeds, secondary, numbr, doubling, certain, threhsold, noticeably, double, observations, along, following, generates, synthetic, them, via, currently, how, nesting, component, variant, issue, current, refactoring, functional, api, update, formatting, local, variables, feature, bugfix, type, tool, needing, follows, guidelines, checklist, measuring, grace, 3461, edited, files, changed, checks, items, requests, dismiss, alert, switched, accounts, appearance, settings, search, pricing, event, talks, workshops, next, partners, company, forum, events, webinars, blog, connect, resources, premium, advanced, ons, trust, evaluate, industries, government, manufacturing, financial, services, healthcare, industry, nonprofits, startups, team, enterprises, stage, businesses, latest, place, developer, articles, sponsors, trending, repos, explore, customize, certifications, scanning, leak, prevention, protection, vulnerability, detection, remediation, secure, cloud, environments, codespaces, analysis, enforcement, automation, ship, coding, agent, tasks, automated, work, hosting, collaboration, platform, menu, skip,
Text of the page (random words):
ic outdated duplicate resolved low quality hide comment code review this pull request introduces a new scaling benchmark for angular build library under scripts benchmarks library builder including fixture generation benchmark execution and integration into the main benchmark cli the review feedback highlights several important improvements resolving node js 18 compatibility issues caused by import meta dirname robustly handling broken symlinks to avoid eexist errors validating cli inputs sizes depths iterations logging process spawn errors and wrapping the benchmark execution in a try finally block to ensure temporary directories are always cleaned up sorry something went wrong uh oh there was an error while loading please reload this page all reactions comment thread scripts benchmarks library builder fixtures mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmark mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page dherges commented oct 9 2026 view reviewed changes dherges left a comment copy link copy markdown author there was a problem hiding this comment choose a reason for hiding this comment the reason will be displayed to describe this comment to others learn more choose a reason spam abuse off topic outdated duplicate resolved low quality hide comment open questions towards the apf references and benchmark design sorry something went wrong uh oh there was an error while loading please reload this page all reactions comment thread scripts benchmark mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder fixtures mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder fixtures mts resolved uh oh there was an error while loading please reload this page dherges commented oct 10 2026 copy link copy markdown author gemini review 1 gemini code assist bot reacted with eyes emoji all reactions 1 reaction sorry something went wrong uh oh there was an error while loading please reload this page gemini code assist bot reviewed oct 10 2026 view reviewed changes gemini code assist bot left a comment copy link copy markdown there was a problem hiding this comment choose a reason for hiding this comment the reason will be displayed to describe this comment to others learn more choose a reason spam abuse off topic outdated duplicate resolved low quality hide comment code review this pull request introduces a new scaling benchmark subsystem library builder to measure cold build times for angular build library across various layouts and style variants feedback on the implementation focuses on improving the robustness of dependency resolution and symlinking in index mts by replacing manual pnpm store parsing with createrequire and require resolve updating existing symlinks if they point to outdated targets and preventing a potential crash in parseintlist in benchmark mts when a single numeric argument is passed sorry something went wrong uh oh there was an error while loading please reload this page all reactions comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmarks library builder index mts resolved uh oh there was an error while loading please reload this page comment thread scripts benchmark mts resolved uh oh there was an error while loading please reload this page dherges marked this pull request as draft october 12 2026 07 45 dherges commented oct 12 2026 view reviewed changes comment thread scripts benchmarks library builder readme md comment on lines 124 to 161 flat layout inline templates only entries median ms entry 10 3 25s 324 63 20 3 28s 164 14 50 3 33s 66 61 100 3 75s 37 53 200 4 31s 21 56 300 4 90s 16 35 400 5 51s 13 78 500 6 18s 12 36 1000 10 29s 10 29 2000 19 47s 9 74 deep layout inline templates only entries median ms entry 16 3 27s 204 47 32 3 30s 103 16 64 3 57s 55 74 128 4 20s 32 84 256 4 70s 18 36 512 6 31s 12 32 1024 10 25s 10 01 2048 21 67s 10 58 conclusion per entry build time decreases monotonically across both layouts and flattens out around 9 6 10 6ms entry at the largest sizes tested no u shaped climb back up at any tested size up to 2000 2048 entries unlike the pattern reported for ng packagr past n 200 500 flat and deep track each other closely at comparable entry point scale suggesting the directory layout itself isn t a meaningful cost driver for the native builder just raw entry point count deep s depth 11 run 2048 entries showed slightly higher run to run variance 20 78s 22 46s than flat s 2000 entry run 19 46s 19 63s possibly directory tree walk overhead at depth but not a regression and not yet investigated further all 60 builds 10 sizes x 3 iterations x 2 layouts completed with status ok zero failures dherges oct 12 2026 copy link copy markdown author there was a problem hiding this comment choose a reason for hiding this comment the reason will be displayed to describe this comment to others learn more choose a reason spam abuse off topic outdated duplicate resolved low quality hide comment iterated the benchmark design and ran the first full default sweep against angular build library flat deep layouts iterations 3 inline template fixtures only no u shape unlike ng packagr s reported climb past 200 500 entries per entry build time decreases monotonically here and flattens around 9 6 10 6ms entry at the top end 2000 entries flat and deep layouts track closely at comparable entry point scale directory structure doesn t appear to be a meaningful cost driver on its own all 60 builds 10 sizes 3 iterations 2 layouts succeeded zero failures sorry something went wrong uh oh there was an error while loading please reload this page all reactions dherges force pushed the test ng build library builder benchmarks branch from fdb8f5a to dd25d8f compare october 16 2026 20 10 dherges commented oct 17 2026 view reviewed changes comment thread scripts benchmark mts comment on lines 209 to 211 style flavor f f style flavor targets one or more of inline inline css inline scss external external css external scss default inline dherges oct 17 2026 copy link copy markdown author there was a problem hiding this comment choose a reason for hiding this comment the reason will be displayed to describe this comment to others learn more choose a reason spam abuse off topic outdated duplicate resolved low quality hide comment updated benchmark design style flavor is an independent sweep axis covering 6 variations of inline vs external template only vs css vs scss sorry something went wrong uh oh there was an error while loading please reload this page all reactions dherges added 12 commits october 23 2026 09 48 test angular build add library builder scaling benchmark suite e8638b9 fixup test angular build add library builder scaling benchmark suite f760cb4 fixup test angular build add library builder scaling benchmark suite 0aaa731 fixup test angular build add library builder scaling benchmark suite 90a5c08 fixup test angular build add library builder scaling benchmark suite fe70c22 fixup test angular build add library builder scaling benchmark suite c7a812b fixup test angular build add library builder scaling benchmark suite 1d7a102 fixup test angular build add library builder scaling benchmark suite 8f7573d recorded reuslts for the iteration 1 baseline sweep against the native libray builder indicating that the new native library builder scales gracefully far far better than ng packagr no u shape curve detected fixup test angular build add library builder scaling benchmark suite 4a9b4ec reported benchmark results in csv file or json fixup test angular build add library builder scaling benchmark suite d48bd74 implemented the style flavor matrix style flavor see style flavor matrix above inline inline css inline scss external external css external scss as a sweep axis independent of layout count depth defaults to inline only matching the prior behavior since the full matrix multiplies build count by up to 6x opt in with style flavor list fixup test angular build add library builder scaling benchmark suite f7e154a fix implemented in harness mts index mts new preparebuilderenv creates once per run idempotently a tiny symlink directory mapping angular build angular devkit core angular devkit architect to their dist output combined with a node_path pointing at that shim dir plus each package s own existing node_modules where ajv formats sass esbuild etc are already correctly pnpm linked this lets the spawned cli resolve every bare require however deeply nested buildonce now takes and applies this nodepath merging with any pre existing node_path rather than clobbering it fixup test angular build add library builder scaling benchmark suite ead4fa9 adds a builder axis library ng packagr so fixtures can be generated in either builder s project shape for a b benchmarking fixtures only no results recorded dherges force pushed the test ng build library builder benchmarks branch from d4d7b63 to ead4fa9 compare october 23 2026 09 49 fixup test angular build add library builder scaling benchmark suite 45eb7d1 report a b bnchmark results from 2026 10 23 test run dherges commented oct 23 2026 view reviewed changes comment thread scripts benchmarks library builder readme md comment on lines 260 to 276 sh pnpm admin benchmark library builder builder library ng packagr layout flat count 10 50 100 200 500 1000 iterations 3 csv path entries library median ng packagr median speedup library ms entry ng packagr ms entry 10 3 33s 5 10s 1 5x 333 2 510 3 50 3 54s 14 20s 4 0x 70 8 283 9 100 3 93s 25 21s 6 4x 39 3 252 1 200 4 47s 48 16s 10 8x 22 3 240 8 500 6 03s 122 58s 20 3x 12 1 245 2 1000 10 01s 269 98s 27 0x 10 0 270 0 conclusion library costs 3s fixed plus 5 8ms per additional entry point so ms entry keeps falling to 10 at n 1000 ng packagr costs 225ms per additional entry up to n 200 then 248ms 200 500 and 295ms 500 1000 the speedup grows from 1 5x n 10 to 27x n 1000 dherges oct 23 2026 copy link copy markdown author there was a problem hiding this comment choose a reason for hiding this comment the reason will be displayed to describe this comment to others learn more choose a reason spam abuse off topic outdated duplicate resolved low quality hide comment with the builder axis added i ran a first a b comparison angular build library vs angular build ng packagr see above the results here indicate that ng packagr scales roughly linear while the library builder achieves a much much higher throughput of entry points subtracting the upfront fixed build time cost the library builder records far better build times per entry point than ng packagr given this fixture set the results for the library builder are in line with the earlier benchmark from 2026 10 12 i then spent some time thinking about the results why is the new library builder so much better i investigated into ng package json and file i o as a possible cause of the bottleneck in ng packagr but i don t think it s the actual cause the glob pattern matching for ng package json is done just once up front i ran the glob pattern in a little js script and it discovered the 1000 entry points in less than 20 ms so that cannot explain the gap with today s knowledhe i think the speedup should be attributed to the different architectures ng packagr has that sequential one entry point at a time approach the native library builder uses a shared compilation host and shared worker pools afaik and i am thinking this is most likely the reason the results indicate that build time per entry point is far better in the native library builder details for reference the log and csv output builder layout styleflavor entrypoints iteration durationms status library flat inline 10 1 3332 00 ok library flat inline 10 2 3365 44 ok library flat inline 10 3 3262 08 ok ng packagr flat inline 10 1 5142 96 ok ng packagr flat inline 10 2 5102 76 ok ng packagr flat inline 10 3 4979 11 ok library flat inline 50 1 3592 71 ok library flat inline 50 2 3539 35 ok library flat inline 50 3 3500 46 ok ng packagr flat inline 50 1 13840 76 ok ng packagr flat inline 50 2 14574 59 ok ng packagr flat inline 50 3 14197 36 ok library flat inline 100 1 3928 30 ok library flat inline 100 2 4721 72 ok library flat inline 100 3 3869 63 ok ng packagr flat inline 100 1 24539 76 ok ng packagr flat inline 100 2 26671 17 ok ng packagr flat inline 100 3 25214 73 ok library flat inline 200 1 4420 91 ok library flat inline 200 2 4783 95 ok library flat inline 200 3 4466 94 ok ng packagr flat inline 200 1 48161 94 ok ng packagr flat inline 200 2 51275 43 ok ng packagr flat inline 200 3 48019 29 ok library flat inline 500 1 6029 87 ok library flat inline 500 2 6038 62 ok library flat inline 500 3 5913 01 ok ng packagr flat inline 500 1 120750 63 ok ng packagr flat inline 500 2 125114 82 ok ng packagr flat inline 500 3 122579 81 ok library flat inline 1000 1 9700 14 ok library flat inline 1000 2 10059 69 ok library flat inline 1000 3 10005 39 ok ng packagr flat inline 1000 1 269976 18 ok ng packagr flat inline 1000 2 274234 54 ok ng packagr flat inline 1000 3 255787 87 ok node no warnings experimentalwarning experimental transform types scripts devkit admin mts benchmark library builder builder library ng packagr layout flat count 10 50 100 200 500 1000 iterations 3 csv tmp ab flat 1792749996 csv library builder library flat 10 inline 10 entry points run 1 3 done in 3 33s run 2 3 done in 3 37s run 3 3 done in 3 26s median 3 33s 333 20ms entry library builder ng packagr flat 10 inline 10 entry points run 1 3 done in 5 14s run 2 3 done in 5 10s run 3 3 done in 4 98s median 5 10s 510 28ms entry library builder library flat 50 inline 50 entry points run 1 3 done in 3 59s run 2 3 done in 3 54s run 3 3 done in 3 50s median 3 54s 70 79ms entry library builder ng packagr flat 50 inline 50 entry points run 1 3 done in 13 84...
|