WEB / WEB-06.03

URL resolution

Follow how one relative reference yields different URLs under different bases; separate parsing success from resource loading.

CORE · RED · BLUE · ARTIFACT · PRACTICE

Read every scenario without motion

Current directory

Reference selected
A link reference exists; there is no absolute result yet.
Reference
notes
Local resolver
Base: pending
Application result
Destination: absent
Base context ready
This operation uses a declared base string and creates no Document.
Reference
notes
Local resolver
https://books.example/catalog/lesson.html?lang=tr#old
Application result
Destination: absent
Resolution is applied
The base current directory is /catalog/, so notes replaces the final segment and produces /catalog/notes. This path reference carries neither the old query nor fragment.
Reference
notes
Local resolver
https://books.example/catalog/notes
Application result
Destination: pending consumption
Final URL is a separate result
The application receives the normalized candidate; no network has started.
Reference
notes
Local resolver
Parse: success
Application result
https://books.example/catalog/notes
The operation boundary remains visible
Address resolution is not resource loading or permission.
Reference
notes
Local resolver
Local calculation complete
Application result
Fetch: not started

Trailing slash

Reference selected
A link reference exists; there is no absolute result yet.
Reference
notes
Local resolver
Base: pending
Application result
Destination: absent
Base context ready
This operation uses a declared base string and creates no Document.
Reference
notes
Local resolver
https://books.example/catalog/lesson/
Application result
Destination: absent
Resolution is applied
The base has a trailing slash, giving current directory /catalog/lesson/. notes resolves within it as /catalog/lesson/notes; no filesystem check occurs.
Reference
notes
Local resolver
https://books.example/catalog/lesson/notes
Application result
Destination: pending consumption
Final URL is a separate result
The application receives the normalized candidate; no network has started.
Reference
notes
Local resolver
Parse: success
Application result
https://books.example/catalog/lesson/notes
The operation boundary remains visible
Address resolution is not resource loading or permission.
Reference
notes
Local resolver
Local calculation complete
Application result
Fetch: not started

Query-only

Reference selected
A link reference exists; there is no absolute result yet.
Reference
?lang=en
Local resolver
Base: pending
Application result
Destination: absent
Base context ready
This operation uses a declared base string and creates no Document.
Reference
?lang=en
Local resolver
https://books.example/catalog/lesson.html?lang=tr#old
Application result
Destination: absent
Resolution is applied
?lang=en retains the base scheme, authority and path, replaces the query and removes the old #old fragment. The result is /catalog/lesson.html?lang=en.
Reference
?lang=en
Local resolver
https://books.example/catalog/lesson.html?lang=en
Application result
Destination: pending consumption
Final URL is a separate result
The application receives the normalized candidate; no network has started.
Reference
?lang=en
Local resolver
Parse: success
Application result
https://books.example/catalog/lesson.html?lang=en
The operation boundary remains visible
Address resolution is not resource loading or permission.
Reference
?lang=en
Local resolver
Local calculation complete
Application result
Fetch: not started

Invalid base

Reference ready
The notes reference requires an absolute context.
Reference
notes
Local resolver
Base: pending
Application result
Destination: absent
Base is invalid
not an absolute base is not a valid base for this constructor.
Reference
notes
Local resolver
Base: invalid
Application result
Destination: absent
TypeError occurs
Failure occurs during parsing; DNS/TLS/HTTP has not started.
Reference
notes
Local resolver
TypeError
Application result
Destination: absent
The failure result is retained
A previous successful URL is not substituted as this operation’s result.
Reference
notes
Local resolver
Parse: failure
Application result
Destination: null; fetch: false

ACTORS · STATE · DATA FLOW

Resolution needs a reference and context

The notes link on Duru’s catalogue page is not an absolute address by itself. Its result depends on processing the reference with a base. Given https://books.example/catalog/lesson.html?lang=tr#old, notes resolves to https://books.example/catalog/notes.

URL resolution differs from DNS resolution. It locally produces an absolute URL; it has not obtained an IP, connected to a server, or read a notes record. A resolved-address observation is distinct from a loaded-resource observation.

Three scene actors show the reference, chosen base, and result. They are roles in a local decision rather than network services, so the steps generate no wire packets.

The final slash changes the current directory

Bases ending in /catalog/lesson.html and /catalog/lesson/ provide different current directories. The former ends its current directory at /catalog/, yielding /catalog/notes. The trailing slash in the latter yields /catalog/lesson/notes.

The parser does not know whether lesson.html is a file. Hierarchical path structure and the trailing slash, not a filename extension, determine the calculation. notes and ./notes use the current directory in these examples.

Concatenating strings can preserve an irrelevant query/fragment or treat the final segment as a directory. The artifact compares both bases using the same reference and records the native constructor’s result.

Follow root and parent references

/notes is root-relative: it retains the base origin context but ignores the base directory. Either catalogue base produces https://books.example/notes. This root is not a server filesystem root.

../notes selects the parent of the current directory. A /catalog/lesson.html base has current directory /catalog/, yielding /notes; /catalog/lesson/ has a deeper current directory and yields /catalog/notes.

These ordinary dot-segment calculations belong to URL path processing and execute no filesystem access. The artifact computes benign catalogue links; selecting an application route remains a separate responsibility.

Query-only and fragment-only references differ

?lang=en selects a new query while retaining the base scheme, authority, and path. Because the new reference carries no fragment, #old is not retained. The result is https://books.example/catalog/lesson.html?lang=en.

#new retains the path and existing query while replacing the fragment: https://books.example/catalog/lesson.html?lang=tr#new. Combine this with the earlier fragment/request-target distinction; resolution itself sends no request.

An absolute https://library.example.org/notes reference selects a destination independent of a valid base’s host/path context. Syntax validity does not establish suitability under the application’s destination policy. The page containing a link does not alone describe its final destination.

Document bases and the failure boundary

An HTML Document’s base commonly comes from its document context; the first base element with href can instead provide the standard’s frozen base URL. document.baseURI is not guaranteed to match the address bar text in every context. about:blank and srcdoc also have fallback rules.

This fixture creates no Document: it passes a declared HTTPS base string to new URL(reference, base). It does not test HTML base inheritance or navigation. The specification defines browser context behavior; the artifact observes the local API.

An invalid base makes the constructor throw TypeError. A failed operation must not display a previous successful destination as its new result: retain no-candidate and fetch-not-started states. Retrying with a valid base starts another calculation rather than rewriting the failure.

Read every scenario without motion

Base context

The reference was selected with its base
notes resolves to /catalog/notes with the lesson.html base, but /catalog/lesson/notes with the /catalog/lesson/ base. One reference differs under different effective bases.
Reference/base evidence
notes / trailing-slash base
URL construction gate
No candidate
Network/access conclusion
No network outcome
The parser-gate outcome was produced
trailing-slash expects https://books.example/catalog/lesson/notes. The parser result does not show which server resources exist.
Reference/base evidence
notes / trailing-slash base
URL construction gate
https://books.example/catalog/lesson/notes
Network/access conclusion
No network outcome
The false-network conclusion was stopped
A candidate proves no request sending, response receipt or exposure. Application destination policy separately evaluates the resolved structured URL.
Reference/base evidence
notes / trailing-slash base
URL construction gate
https://books.example/catalog/lesson/notes
Network/access conclusion
Candidate only / delivery unknown
Query and fragment effects were separated
?lang=en changes the query and removes the old fragment; #new preserves query lang=tr. Final-URL component differences do not measure server locale behavior.
Reference/base evidence
Query changes / fragment retains query
URL construction gate
https://books.example/catalog/lesson/notes
Network/access conclusion
Candidate only / delivery unknown
Root and parent depend on context
/notes resolves at the origin root; ../notes reaches root from lesson.html but /catalog/notes from a trailing directory base. These are local URL-resolution rules, not filesystem access.
Reference/base evidence
Query changes / fragment retains query
URL construction gate
Root / parent require selected base
Network/access conclusion
Candidate only / delivery unknown
The closed-gate or candidate scope was preserved
The resolved href agrees with its selected base context. Link calculation establishes neither same-origin policy nor resource authorization.
Reference/base evidence
Query changes / fragment retains query
URL construction gate
Root / parent require selected base
Network/access conclusion
Resolved candidate / access unknown

Invalid base

The reference was selected with its base
Relative notes is paired with not an absolute base. A reference alone produces no destination; the caller’s selected context must be valid.
Reference/base evidence
notes / invalid base
URL construction gate
No candidate
Network/access conclusion
No network outcome
The parser-gate outcome was produced
URL construction with an invalid base produces TypeError. Destination stays null and fetch never starts. A previous successful case’s href is not carried into the failure branch as a candidate.
Reference/base evidence
notes / invalid base
URL construction gate
TypeError / candidate=null
Network/access conclusion
No network outcome
The false-network conclusion was stopped
TypeError indicates a local URL-construction failure, not HTTP 404 or a transport timeout. Because no destination is produced, this case does not establish that the browser sent a request.
Reference/base evidence
notes / invalid base
URL construction gate
TypeError / candidate=null
Network/access conclusion
No candidate / no fetch
The error belongs to one input pair
Reference notes and base not an absolute base remain in one operation record. Importing another case’s valid base would change the failed construction’s inputs rather than explain its error.
Reference/base evidence
notes / invalid-base record
URL construction gate
TypeError / candidate=null
Network/access conclusion
No candidate / no fetch
Candidate absence was preserved
Parser failure remains TypeError with candidate=null. The construction error is returned to the caller instead of reusing a prior successful href; this branch creates no destination or request.
Reference/base evidence
notes / invalid-base record
URL construction gate
TypeError / candidate=null
Network/access conclusion
No candidate / no fetch
The closed-gate or candidate scope was preserved
Invalid base stops at the local gate with no stale href. The error is not moved to networking or server policy; the needed evidence is reference/base/parser error.
Reference/base evidence
notes / invalid-base record
URL construction gate
TypeError / candidate=null
Network/access conclusion
Local TypeError / fetch not started

ASSUMPTION · BOUNDARY · CONTROL

Identify the context that produced the final URL

Reporting a reference alone omits context: notes can produce different paths under different bases. Record the reference, declared/effective base, and normalized final URL together.

Compare the trailing-slash branch. A different result does not imply inconsistent parsing; its context changed. Without actual fetch or server evidence, address calculations do not establish resource exposure.

Compare query-only and fragment-only cases: ?lang=en changes the query and drops the old fragment, while #new retains lang=tr. Naming the component changed is stronger mechanism evidence than saying the link changed. Server locale behavior has not been measured here.

Do not label local failure as network failure

TypeError from an invalid base is neither a transport timeout nor HTTP404. Identify the incomplete stage: without a parsed result there is no destination for request construction.

Origin/site comparisons or application destination checks use the resolved structured URL. A reference alone does not measure that boundary, and URL resolution is not authorization.

The failed branch keeps destination null. Carrying a previous successful case’s href into it would invent a result for an operation that produced none. In a review, error type, selected base, and candidate presence must belong to the same operation record rather than being assembled from different cases.

Read every scenario without motion

Base change

The operation record has two inputs
Each comparison retains reference and base. current-directory notes with base=https://books.example/catalog/lesson.html?lang=tr#old gives /catalog/notes.
Resolution input evidence
Reference + base same operation
URL comparison
No URL calculation
Observation boundary
No evidence scope
Trailing-slash output was compared
With base https://books.example/catalog/lesson/, the same notes gives /catalog/lesson/notes. The difference is changed base context, not parser inconsistency or a need for manual concatenation.
Resolution input evidence
Reference + base same operation
URL comparison
lesson.html → /catalog/notes; lesson/ → /catalog/lesson/notes
Observation boundary
No evidence scope
Root and parent are compared under the same base
/notes resolves at the origin root. ../notes produces /notes with the lesson.html base and /catalog/notes with the lesson/ base. These are separate rows; retaining reference and base together identifies the directory rule behind each result.
Resolution input evidence
Reference /notes or ../notes; base retained
URL comparison
Root /notes; parent depends on base
Observation boundary
Local URL result; no filesystem access
An absolute reference can select another authority
Absolute reference=https://library.example.org/notes replaces the base with its own URL. Parser output can show an origin change but does not measure application allowance of that destination.
Resolution input evidence
Separate absolute-reference comparison
URL comparison
https://library.example.org/notes
Observation boundary
No evidence scope
The producer and inputs match the record
The href output is produced by the local URL API. Bind reference, base and expected result to the same operation; do not copy a successful href from another base. Document-base inheritance or navigation evidence requires an actual Document.
Resolution input evidence
Reference + base + output retained
URL comparison
Producer: local URL API
Observation boundary
Coverage: declared-string resolution
The changed base explains the result
The differing paths for notes under two bases result from changed context. A URL-resolution record is neither DNS, HTTP receipt nor resource authorization evidence. Each of those questions requires its own producer.
Resolution input evidence
notes / two separate base contexts
URL comparison
lesson.html → /catalog/notes; lesson/ → /catalog/lesson/notes
Observation boundary
Resolution known; network unknown

Query / fragment

Two references were paired with the same base
The supplied base is https://books.example/catalog/lesson.html?lang=tr#old. ?lang=en and #new are separate inputs, compared while the base remains unchanged. No request is made beyond the URL calculation.
Resolution input evidence
Same base; ?lang=en / #new
URL comparison
Component results pending
Observation boundary
No network observation
Query-only retains the path
?lang=en retains scheme/authority and path /catalog/lesson.html. The new query is lang=en and the old #old is removed. This row gives expectedHref=https://books.example/catalog/lesson.html?lang=en.
Resolution input evidence
Reference ?lang=en
URL comparison
Same path; query en; no hash
Observation boundary
No server-locale outcome
Fragment-only retains the existing query
The separate #new row retains the same base path and query lang=tr, replacing the fragment with new. It yields https://books.example/catalog/lesson.html?lang=tr#new. The query-only row’s lang=en is not carried over.
Resolution input evidence
Reference #new; same base
URL comparison
Same path; query tr; hash #new
Observation boundary
DOM navigation unmeasured
The request target and fragment differ
For the latter final URL, the ordinary HTTP path/query candidate is /catalog/lesson.html?lang=tr. #new remains a separate client component. Derivation neither sends a request nor shows that the fragment caused UI scrolling.
Resolution input evidence
Path/query candidate; #new separate
URL comparison
Query tr retained; fragment absent from target
Observation boundary
HTTP and scrolling unobserved
The parser produces the component evidence
Href, search and hash outputs come from the local URL API. Actual Document/baseURI, browser fragment handling and server locale behavior have different producers. A supplied base string does not establish execution of those contexts.
Resolution input evidence
Both references and the same base retained
URL comparison
Producer: URL API; no Document created
Observation boundary
Coverage: component resolution
Results are not filled from another row
Query-only and fragment-only results remain tied to their own inputs. The invalid-base comparison instead gives TypeError, destination=null and fetchStarted=false. No successful href fills the failed row’s destination.
Resolution input evidence
Query en/no hash; query tr/#new
URL comparison
Successful href and TypeError are separate rows
Observation boundary
Component results supplied; no network outcome

IMPLEMENTATION · SECURITY CONTROL · VERIFICATION

Treat the base as explicit context data

Make the base source explicit in code using relative references: configuration, Document context, or caller input. Exercise a fixture set with trailing/non-trailing slash bases and root/query/fragment forms.

Use a standard API such as new URL instead of duplicating parsing through concatenation. Apply the application’s destination requirements to the normalized result and do not continue using an old result after parsing fails.

In the URL lab, predict notes under both bases before comparing command output. If the results differ, reread the base context without changing the reference. This separates changed inputs when diagnosing a bug; it can be a local calculation issue requiring neither a network retry nor a server change.

Attach evidence to the correct stage

A resolution record is produced by the URL parser: compare href or error type for the same reference and base. Establishing a Document base requires actual Document context; establishing resource loading requires fetch/response observations. A local parser result is not evidence of those other stages.

A benign regression table records reference, base, expected href, and error type. Use a separate case when the base changes, and verify that failure leaves no candidate in the model. Keep the calculation visible in readable steps when motion is disabled.

For example, notes gives /catalog/notes with lesson.html but /catalog/lesson/notes with lesson/. Attach query-only and fragment-only results to the same base and identify the changed component. An invalid base retains TypeError alongside destination=null. A successful address calculation establishes neither resource existence nor user access.

NOW APPLY IT TO THE CASE

How are local candidates, parser errors and network outcomes kept separate?

Reference notes chooses different paths with lesson.html and lesson/ bases. An invalid base yields TypeError and a null destination. That local parsing result is neither HTTP 404 nor a network timeout.

Identify the case’s browser and API through URL components. Target, origin and site answer different questions; identify which decision uses each one.

Open the related case file ↗

READ THE EXAMPLE TOGETHER

Relative-reference / base / result ledger

Select a field and examine what it supports and where the evidence ends.

Read the raw artifact and field explanations

{
  "cases": [
    {
      "case": "current-directory",
      "reference": "notes",
      "base": "https://books.example/catalog/lesson.html?lang=tr#old",
      "expectedHref": "https://books.example/catalog/notes"
    },
    {
      "case": "dot-current",
      "reference": "./notes",
      "base": "https://books.example/catalog/lesson.html?lang=tr#old",
      "expectedHref": "https://books.example/catalog/notes"
    },
    {
      "case": "trailing-slash",
      "reference": "notes",
      "base": "https://books.example/catalog/lesson/",
      "expectedHref": "https://books.example/catalog/lesson/notes"
    },
    {
      "case": "root",
      "reference": "/notes",
      "base": "https://books.example/catalog/lesson.html?lang=tr#old",
      "expectedHref": "https://books.example/notes"
    },
    {
      "case": "parent",
      "reference": "../notes",
      "base": "https://books.example/catalog/lesson.html?lang=tr#old",
      "expectedHref": "https://books.example/notes"
    },
    {
      "case": "parent-directory",
      "reference": "../notes",
      "base": "https://books.example/catalog/lesson/",
      "expectedHref": "https://books.example/catalog/notes"
    },
    {
      "case": "query",
      "reference": "?lang=en",
      "base": "https://books.example/catalog/lesson.html?lang=tr#old",
      "expectedHref": "https://books.example/catalog/lesson.html?lang=en"
    },
    {
      "case": "fragment",
      "reference": "#new",
      "base": "https://books.example/catalog/lesson.html?lang=tr#old",
      "expectedHref": "https://books.example/catalog/lesson.html?lang=tr#new"
    },
    {
      "case": "absolute",
      "reference": "https://library.example.org/notes",
      "base": "https://books.example/catalog/lesson.html?lang=tr#old",
      "expectedHref": "https://library.example.org/notes"
    }
  ],
  "invalidBase": {
    "reference": "notes",
    "base": "not an absolute base",
    "expectedError": "TypeError",
    "destination": null,
    "fetchStarted": false
  },
  "documentBaseModel": "Declared strings only; no Document/base element created",
  "networkRequestsExecuted": 0
}

cases.reference: Ordinary link reference to resolve.
cases.base: Declared absolute context for this operation.
cases.expectedHref: Serialization to check against the native URL constructor.
invalidBase.destination: Null after failure; no fallback to a previous success.

Synthetic ordinary links resolved with native Node URL. No Document base inheritance, browser navigation, DNS, or fetch executes.

Two reality checks

Relative resolution concatenates the base and reference.

Path, trailing slash, query, and fragment rules transform components separately. Use a standard parser.

A resolved URL proves successful DNS and resource loading.

Resolution is local calculation; DNS, TLS, HTTP, and application outcomes require separate evidence.

YOUR TURN

Work through this story.

Predict the outcome, find the evidence boundary, and explain it in your own words.

Does the reference notes yield the same URL with bases /catalog/lesson.html and /catalog/lesson/?

No; the results are /catalog/notes and /catalog/lesson/notes.
Yes; appending notes to the string is sufficient.

Read the explained answer

No; the results are /catalog/notes and /catalog/lesson/notes. — The final segment of the file-like base is replaced; the trailing-slash base resolves within its directory.

Which different components are preserved by query-only and fragment-only references?

Selected lesson artifact fields; teaching example

cases[6]: {"case":"query","reference":"?lang=en","base":"https://books.example/catalog/lesson.html?lang=tr#old","expectedHref":"https://books.example/catalog/lesson.html?lang=en"}
cases[7]: {"case":"fragment","reference":"#new","base":"https://books.example/catalog/lesson.html?lang=tr#old","expectedHref":"https://books.example/catalog/lesson.html?lang=tr#new"}

Both reset the path to the root.
?lang=en preserves the path, replaces the query and removes the old hash; #new preserves the query.

Read the explained answer

?lang=en preserves the path, replaces the query and removes the old hash; #new preserves the query. — ExpectedHref explicitly shows different inherited components in the two resolutions.

Why is TypeError from an invalid base not an observed network-request failure? Also distinguish Document base context.

Read the explained answer

An invalid absolute base produces no destination; selected fetchStarted=false. This local URL API error is not a remote server response. The fixture uses explicit base strings and creates no actual Document/base element. NetworkRequestsExecuted=0 prevents reachability claims from parsing outcomes.

□ I separated local URL error from network failure.

□ I distinguished explicit base strings from Document context.

Open the application console · request / response / browser

Also open request, URL, cookie and DOM examples

MANUAL LAB

URL Boundary Lab

Predict components, origin/site comparisons, and relative resolution before comparing native API output.

Open the prediction → observation → explanation workbook ↗Source on GitHub ↗

Sources and scope

WHATWG — URL StandardGoals · 4.1 URLs · 4.2 URL miscellaneous: default port · 4.4 URL parsing · 4.7 Origin · 6.1 URL class
RFC 3986 — URI Generic Syntax1.1.3 URI, URL, and URN · 2.1 Percent-Encoding · 2.4 When to Encode or Decode · 3 Syntax Components · 5.2 Relative Resolution
MDN — Resolving relative referencesCurrent directory relative · Parent-directory relative · Root relative
MDN — URL constructorSyntax · Parameters · Exceptions
WHATWG HTML — Parsing URLs and document base URLs2.4.2 Parsing URLs · 2.4.3 Document base URLs

YOUR NOTES / THIS BROWSER

URL resolution

1. Core · Mechanism

Current directory

Trailing slash

Query-only

Invalid base

Resolution needs a reference and context

The final slash changes the current directory

Follow root and parent references

Query-only and fragment-only references differ

Document bases and the failure boundary

2. Red · Trust & Impact

Base context

Invalid base

Identify the context that produced the final URL

Do not label local failure as network failure

3. Blue · Secure development & defense

Base change

Query / fragment

Treat the base as explicit context data

Attach evidence to the correct stage

How are local candidates, parser errors and network outcomes kept separate?

Two reality checks

Does the reference notes yield the same URL with bases /catalog/lesson.html and /catalog/lesson/?

Which different components are preserved by query-only and fragment-only references?

Why is TypeError from an invalid base not an observed network-request failure? Also distinguish Document base context.

URL Boundary Lab

Sources and scope