XmlEncoder: ordering of elements #17

Ma-He · 2018-12-15T17:44:43Z

Hi,
is there any possibility to set the order of attributes in an element while encoding a model-instance?
I'm trying to build a GPX reader/writer with this little awesome framework and have noticed that the order of my attributes in an element varies from encode to encode. Am I missing something?

Thanks in advance

MaxDesiatov · 2018-12-15T19:18:19Z

Hi @Ma-He, thank you reporting this issue. There's a .sortedKeys option that can be set for XMLEncoder.outputFormatting property.

We plan to publish auto-generated docs for XMLCoder soon, which would make this option more visible to users.

Also, if you're using this option, you might need to switch to master branch as it was recently reported that this option was broken. This was fixed in #11 and #14 and merged to master.

regexident · 2018-12-15T20:00:02Z

Sorting is not was OP is looking for. Ordering is. That is <foo>…</foo> has to appear before <bar>…</bar>, if .foo was encoded before .bar.

This is not supported currently as XMLCoder uses unordered NSDictionary in its internal keyed encoding container.

MaxDesiatov · 2018-12-15T20:03:55Z

Thanks @regexident, I understand now. Still, my understanding was that the main concern is about determinism. I hope that .sortedKeys is able to solve that. I guess we have to transition to a tuple array for internal storage to maintain order as well.

regexident · 2018-12-15T20:09:28Z

I'm fairly certain as writing a Codable-powered GPX implementation was my reason for dabbling with XMLCoder (or rather it's predecessor XMLParser) as well.

GPX (like many other XML schemas) expects certain keys to have a specific order:

Search for  in the GPX 1.1 Schema.

This requirement of specific order also applies to elements, not just attributes.

Ma-He · 2018-12-16T09:21:53Z

Hey guys,

thanks for your quick responses. As @regexident said, ordering it is what I`m looking for :-)

) This PR removes any remaining use of `NS…Array`/`NS…Dictionary`/`NSNumber`/`NSDecimalNumber`/`NSNull`/… (as storage) from the code-base. It introduces an internal type … ```swift internal enum Box { case null(NullBox) case bool(BoolBox) case decimal(DecimalBox) case signedInteger(SignedIntegerBox) case unsignedInteger(UnsignedIntegerBox) case floatingPoint(FloatingPointBox) case string(StringBox) case array(ArrayBox) case dictionary(DictionaryBox) } ``` … as well as accompanying variant box types, replacing use of `Any`/`NS…`. 👷🏻‍♀️It improves type-safety by reducing use of `Any` (from 60 down to 19 occurrences) as well as `NSObject` (from 37 down to 1 occurrence). 🏗It further more levels the field for improvements/additions such as [support for order-preserving elements & attributes](#17). 💡Thanks to encapsulation we can now safely change the inner logic of `DictionaryBox` to retain insertion order. **Edit**: We ended up replacing aforementioned `enum Box` with a protocol: ```swift protocol Box { var isNull: Bool { get } var isFragment: Bool { get } func xmlString() -> String? } ``` * Added basic box types * Migrated decoding logic to proper boxes * Migrated encoding logic to proper boxes * Moved auxiliary types into their own dedicated files and group * Replaced use of `.description` with explicit dedicated `.xmlString` * Made box types conform to `Equatable` * Added basic unit tests for box types * Removed manual `Equatable` conformance, changed fragment boxes to structs * Removed redundant use of `internal` on internal types * Simplified logic of `func createElement` instances * Removed `enum Box` in favor of `protocol Box` * Fixed error description for `.typeMismatch` * Shortened type names of `FloatingPointBox`, `SignedIntegerBox` and `UnsignedIntegerBox` * Moved unboxing logic into dedicated Box initializers * Removed last remaining use of explicit `internal` related to this PR * Changed `Unboxed` types of `IntBox` & `UIntBox` to explicit 64bit types * Made `.xmlString` conform to W3.org’s XML-spec * Renamed `.init(string:…)` to `.init(xmlString:…)` * Implemented conformity to encoding strategies * Simplified box unit tests * Added minimalist decoding unit tests for individual box types * Fixed bug related to `Float` encoding and simplified internal encoding/decoding locig

regexident · 2018-12-22T21:36:29Z

Nitpick: It's not the order of attributes that's unsupported, it's the order of elements.

The XML specification explicitly defines the order of attributes to be of no importance:

The Name in the start- and end-tags gives the element's type. [Definition: The Name-AttValue pairs are referred to as the attribute specifications of the element], [Definition: with the Name in each pair referred to as the attribute name ] and [Definition: the content of the AttValue (the text between the ' or " delimiters) as the attribute value.] Note that the order of attribute specifications in a start-tag or empty-element tag is not significant. (emphasis mine) Source

As such NSXMLParser (which XMLCoder is based on) doesn't even provide a way to retrieve an element's attributes in the order of appearance.

MaxDesiatov · 2018-12-23T09:16:19Z

TLDR: We're slowly working towards ordered child elements, but not quite sure if ordering of attribues will be possible in the current implementation.

@Ma-He could you please clarify if you're looking for ordering of XML attributes or child elements?

MaxDesiatov · 2018-12-23T09:17:24Z

If you're specifically interested in supporting GPX, looks like it only requires ordered elements, not attributes.

Ma-He · 2019-01-04T10:48:18Z

@MaxDesiatov you`re right, just the elements

Unifies the representation of `T` where `T: Box` and makes it explicit where shared reference semantics is expected. With this code performance is consistently 5-30% worse, could this be caused by copies of value types? Can this be improved when Swift 5.0 is released or whatever the version is that allows us to improve performance of value types? After reviewing related issues we'd prefer XMLCoder to be slightly slower, but with #12, #17 and #25 fixed than the opposite 😞 * Added `SharedBox` type for on-demand reference semantics * Remove `print` that kills performance tests * Handle all cases with SharedBox containers * Add RJITest back to the project file, fix the test * Remove unused internal initializer

CineDev · 2019-02-12T17:15:18Z

I have to say you doing great job!
But current iteration of XMLEncoder is basically broken for use cases where the ordering is important. I tried to use it in my simple text editor. I'm storing each paragraph individually in my model. But from save to save my model xml representation gets shuffled, so the text information in the saved document become completely useless.
I hope, the ordering is on the way.

Resolves #17 * First attempt at ordered encoding * Fix line length linter warning * Rename DynamicNodeEncoding function, fix order

MaxDesiatov · 2019-03-01T11:14:08Z

@Ma-He @CineDev this is now fixed in master, no source-breaking API changes, should work out of the box. .sortedKeys now only has effect on attributes, not attributes and elements like before.

Add benchmark baselines Pull in tests (CoreOffice#11) Add Decoding support for choice elements (CoreOffice#15) Fix indentation (CoreOffice#16) Replace usage of XCTUnrwap (CoreOffice#19) Add nested choice tests (CoreOffice#18) Add falsely passing double array roundtrip test (CoreOffice#17)

MaxDesiatov assigned hodovani Dec 15, 2018

regexident mentioned this issue Dec 18, 2018

Overhaul internal representation, replacing NS… with …Box types #19

Merged

jsbean mentioned this issue Dec 22, 2018

[RFC] Add OrderedDictionary #37

Closed

MaxDesiatov mentioned this issue Jan 12, 2019

Keyed/unkeyed boxes as structs #36

Merged

MaxDesiatov unassigned hodovani Jan 19, 2019

MaxDesiatov changed the title ~~XmlEncoder: ordering of attributes in an element~~ XmlEncoder: ordering of elements Feb 27, 2019

MaxDesiatov self-assigned this Feb 27, 2019

MaxDesiatov mentioned this issue Feb 27, 2019

Ordered encoding #82

Merged

MaxDesiatov closed this as completed in #82 Mar 1, 2019

MaxDesiatov added a commit that referenced this issue Mar 1, 2019

Ordered encoding (#82)

f8616cf

Resolves #17 * First attempt at ordered encoding * Fix line length linter warning * Rename DynamicNodeEncoding function, fix order

bwetherfield added a commit to bwetherfield/XMLCoder that referenced this issue Jul 17, 2019

Add falsely passing double array roundtrip test (CoreOffice#17)

909ddee

bwetherfield added a commit to bwetherfield/XMLCoder that referenced this issue Jul 26, 2019

Add falsely passing double array roundtrip test (CoreOffice#17)

597c97f

bwetherfield added a commit to bwetherfield/XMLCoder that referenced this issue Jul 27, 2019

Add falsely passing double array roundtrip test (CoreOffice#17)

a01c4e7

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

XmlEncoder: ordering of elements #17

XmlEncoder: ordering of elements #17

Ma-He commented Dec 15, 2018

MaxDesiatov commented Dec 15, 2018

regexident commented Dec 15, 2018

MaxDesiatov commented Dec 15, 2018

regexident commented Dec 15, 2018 •

edited

Ma-He commented Dec 16, 2018

regexident commented Dec 22, 2018 •

edited

MaxDesiatov commented Dec 23, 2018 •

edited

MaxDesiatov commented Dec 23, 2018

Ma-He commented Jan 4, 2019

CineDev commented Feb 12, 2019 •

edited

MaxDesiatov commented Mar 1, 2019 •

edited

XmlEncoder: ordering of elements #17

XmlEncoder: ordering of elements #17

Comments

Ma-He commented Dec 15, 2018

MaxDesiatov commented Dec 15, 2018

regexident commented Dec 15, 2018

MaxDesiatov commented Dec 15, 2018

regexident commented Dec 15, 2018 • edited

Ma-He commented Dec 16, 2018

regexident commented Dec 22, 2018 • edited

MaxDesiatov commented Dec 23, 2018 • edited

MaxDesiatov commented Dec 23, 2018

Ma-He commented Jan 4, 2019

CineDev commented Feb 12, 2019 • edited

MaxDesiatov commented Mar 1, 2019 • edited

regexident commented Dec 15, 2018 •

edited

regexident commented Dec 22, 2018 •

edited

MaxDesiatov commented Dec 23, 2018 •

edited

CineDev commented Feb 12, 2019 •

edited

MaxDesiatov commented Mar 1, 2019 •

edited