ID
|
Status
|
Spec
|
Topic
|
Class
|
Title
|
Summary
|
Proposed by
|
Original Proposal
|
Discussion History
|
1
|
Closed
|
1.1 Spec
|
Extensibility
|
Design
|
Extending Attribute Values
|
27 Jan 03 – Updated in 1.1 Draft 2d
10 December 2002 = Unanimous vote to use the 'x-'
mechanism for attribute value extension. .
Proposal for validation mechanism for XLIFF files that have been
customized. Three proposed alternatives are proposed for the schema:
1. From Yves uses the 'union'mechanism
2. a new proposal from Christian which uses the 'redefine' mechanism. The
argument in favor of the 'redefine' mechanism is that it provides a way of
validating customized values and is more flexible, The argument for the
'union' mechanism is that it is simple and more easily implemented.
3. Shigemichi's "x-" proposal: to use a TMX style
prefix for any custom attribute. Danger here is that custom
extensions for commonly named attributes ie, "x-button",
could result in ambiguity or worse - invalid identification. Sugestion to
extend with additional namespace identifier was not met with some
resistance.
|
Christian Lieske
|
Christian's
Proposal
|
sec
2.4 in WIP 1.1 spec
|
2
|
Closed
|
1.1 Spec
|
Interoperability
|
Design
|
Context Group
|
27 Jan 03 – Updated in 1.1 Draft 2d
3 Dec - unanimous vote to remove from the spec.
Original:
Further action requried: remove from the spec.
XLIFF 1.1 Working Draft, Section 2.3, paragraph 2 talks about the
context-group element. In that section it talks about the different
purposes for the context information, i.e. TMs, translators, etc. The final
sentence refers to using PIs to indicate the different purposes. However,
we are no longer specifying the use of PIs and we have never enumerated the
purposes of context information. .
|
John Reid
|
John
Reid's Proposal
|
Mark
Levins' Suggestion
|
3
|
Closed
|
1.1 Spec
|
Interoperability
|
Design
|
New elements
"default" and "defaults"
|
27 Jan 03 – Updated in 1.1 Draft 2d
21 Jan 2003 No further attributes submitted at meeting –
resolution stands as defined by John and now update to specification
required.
14 Jan 2003: This
was accepted in a vote, but additional default attributes might be added to
the <group> element at next week's meeting.
7 Jan 20003 Members of the TC have been asked express
opinions on this before next week when there will be a vote.
19 Dec 2002: the proposal is to add these attributes to
group for the express purpose of setting defaults for all child
<trans-unit>s: charclass, maxbytes, maxheight, maxwidth, minbytes,
minheight, minwidth, size-unit, translate, reformat
The attributes of the <trans-unit> that have not been considered for
the <group> element are listed below: approved, phase-name, tu-state
(proposed)
17 Dec 2002: John Reid to draft proposal for extending <group> for
storing default attributes.
10 Dec 2002 : Mark raised John Reid's suggestion of using <group> to
store defaulted values and adding the defaultable attributes to the group
element. Tony suggested closing out on the original proposal and not using
XPath by replacing the default strategy with <group>. John Reid volunteered
to redefine the proposal around new attributes of <group> for the
next meeting.
Original:
Amended Requirements:
R.1 a mechanism to allow defaulting for XLIFF data categories
R.2 formal representation of data category is secondary (i.e. the mechanism
should be applicable to attributes and elements)
R.3 mechanism should work for all XLIFF data categories
R.4 location for defaulting information is secondary (i.e. default in
central location, default at specific attributes or elements, and default
at all attributes and elements is acceptable)
R.5 XPath should not be used to relate default settings to the elements or
attributes to which they pertain (let's call this 'target')These
requirements boil down to 3 questions:
Q.1 What is defaulted?
Q.2 How is it defaulted?
Q.3 Where is it defaulted?
Originally submitted proposal (which did not meet R.5), answered the
questions as follows:
P1.A1 allow defaulting for any XLIFF data category
P1.A2 use XPath to designate the targets for default settings
P1.A3 use a new central element 'defaults'
Amended proposals which take into account R.5 :
P1': like P1 but without XPath
The idea here is that each target explicitly names the defaults which
should be used for it. From my understanding, this is not really kosher, since
for example the way to identify relationships (or 'targets')is a
proprietary and not very efficient one. XPath is the standard for this.
Accordingly, I would ask the TC members to reconsider my original proposal.
P2: defaults are encoded at the level of the 'group' element (John's
proposal)
P3: defaults are encoded 'in the vicinity'of the XLIFF element to which
they pertain (Mark's proposal)
todo:
a)define defaultable data categories Q.1
b)design a representation for default settings (Q.2); this has include a
way to identify to which XLIFF data category a setting pertains
|
Christian Lieske
Amended by John Reid
|
Christian's
Proposal
|
Christian's
Amended
Proposal without XPATH
(29
Dec) John Reid’s use of <group> attributes proposal
|
4
|
Assigned to Editor
|
1.1 Spec
|
Interoperability
|
Design
|
Phase names in Alt Trans
|
10 Dec 2002: John outlined his most recent proposal
which required very little changes to the XLIFF specification on the basis
that it already provided enough support for tracking phases and histories.
Tony motioned for voting on John's proposal at the next meeting, giving
everyone enough time to digest the proposal and be happy with the suggested
additional attribute values.
Proposal:
Since the current phase information can be retrieved from the
<target> attribute I think we don't need phase-name attribute in the
<trans-unit> element and my proposal is to remove it from there.
Mark's additional suggestion for "reason" attribute:
Provide another required attribute in <alt-trans>
"reason" to indicate the reason why a given <alt-trans> is
an alternate translation. A few suggested values for such an
attribute are 'TM Suggestion', 'MT Suggestion', 'Rejected-Inaccurate',
'Rejected-Spelling', 'Rejected-Grammar' and 'Rejected-Length'.
List would remain open, but with a list of suggeted values.
|
Mirek Driml
|
Mirek's
proposal
|
05
Dec- Mirek's Clarification
05 Dec - Mark's suggestion
adding "reason"
John
Reid's Proposal (10 Dec)
|
5 |
Closed
|
1.1 Spec
|
Editorial
|
Editorial
|
"Zero, One or More" Language
|
5 Dec 02 After a vote it was
unanimously agreed that this issue is at the discretion of the
specification editor and to leave spec unchanged.
Propose replacing the phrase
"zero, one or more" with the more precise "zero or
more." This should be done globally
|
Bryan Schnabel
|
|
|
6 |
Assigned to Editor
|
1.1 Spec
|
Interoperability
|
Design
|
reformat Element revisited
|
27
January Option 5 was selected in a vote.
The
result details were: option 4, 1 vote; option 5, 11 votes; 1 abstain.
27
January 2003 – Vote will take place at 27 January 2003 meeting. Voting will be limited to these 6
options:
Option
1: Siblings
Option
2: Restructure
Option
3: Embedded
Option
4: Combined
Option
5: Extended
Option
6: Do nothing - defer to future release
Chair
recommends Option 5
24
Jan 2003 - Option 5, submitted by
Doug
Extend
the possible values for the 'reformat' attribute to provide sufficient
control. XLIFF 1.0 presently uses ";"-delimited lists within
attribute values to store multiple values. The 'coord' attribute is an
example. It's value is actually four: "x;y;cx;cy", where
"#" can be used for 'don't care'.
So
let's extend 'reformat' the same way. Of course, we keep "yes"
and "no" for compatibility.
"yes" = all
format attributes may be changed
"no" = no
format attributes may be changed
...or a semicolon-delimited list of the following in
any order. If an attribute is listed, it means it may be reformatted.
coord = all 4 coords
coord-x
coord-y
coord-cx
coord-cy
font = all 3 font values
font-name
font-size
font-weight
css-style
style
exstyle
Example,
<trans-unit
coord="#;#;183;272" font="Arial;2;normal"
reformat="coord-cx;font-name" ...>
<source>...</source>
<target
coord="#;#;181;272"
font="System;2;normal">...</target>
<alt-trans
coord="#;#;183;272" font="Arial;2;normal">
<target
coord="#;#;180;272" font="Arial
Bold;2;normal">...</target>
<target
coord="#;#;185;272" font="Arial, Helvetica;2;normal">...</target>
</alt-tran>
</trans-unit>
Parsing
the reformat list is fairly easy, even with XSLT, which has a limited set
of string functions.
This
option is 100% compatible, both forward and backward. It does not affect
the structure at all. The only problem I can foresee an XLIFF 1.0 tool
having is if an invalid value for reformat is assumed to be "yes"
instead of "no" and allows some values to be changed that should.
That is, an XLIFF 1.0 tool could interpret a value of "coord-cx;font-name"
as "no" and not allow any of the format value to change. Of
course, if it assumed "no" instead of "yes" it would
not allow any changes. Since the default value for 'reformat' is
"yes", I don't see either of the possibilities as being too harmful.
21
Jan 2003 More discussion regarding the 4 options, but with informal “straw
poll” indicating overwhelming support for “Option 2 – Restructure”. However, it was noted that this option does not preserve backwards
compatibility with 1.0.
14
Jan 2003 There was much discussion over the benefits of each of Matt's
proposals (in essence, sibling elements to <source> and
<target>, or child elements to <source> and <target>)
with various advantages of each option pointed out, from backwards
compatibility to neatness, to shortcomings with <alt-trans> and
workarounds for these. We will
discuss this for one more week and vote at the next meeting on whether to
accept this proposal in concept.
7
Jan 2003 Mat outlined his proposal (see discussion history) This will be
discussed on 14 Jan 2003
17 Dec 02: Mat to rewrite proposal in full as it would appear in the
specification, and group to respond with any questions or
issues. This will be done before the next meeting.
10 Dec 02: Matt tabled new suggestions for reformatting based on
previously sent email. Mark raised objections to instruction-based reformat
element that would require similar functionality to XPath and suggested
adding new specific elements for content that can be changed as part of the
translation process (e.g. font, coord, style etc) where these elements
could contain a boolean attribute to indicate whether they could be
altered. Matt agreed to further investigate this approach and create some
examples for the next meeting. Enda then raised the question of how this
would affect the 'default' discussion and Matt brought up the ability to
default a translation or translatable content. Matt agreed to try to factor
these two points into his investigation for the next meeting.
Simple, non-verbose, mechanisms to:
1. Indicate the translatability of any attribute/element, or XLIFF
standard values or extensions.
2. Store source and translated values for any structure marked as
translatable
Proposals
1) A closed list of XLIFF standard attributes and elements
that may be modified during translation. E.G state, target text
2) Each member of the list will either have before/after
placeholders or will be simply updated without keeping previous values
3) No other attribute/element may be translated unless
specifically marked as translatable
4) Provide place holders for any modified element
|
Mat Lovatt
|
Mat's
Initial Proposal
Mat's
Revised Proposal
Mat's
final proposal (7 Jan)
Mat’s
summary of Options 1 – 4; to be used in ballot (submitted 23 Jan 2003)
Doug’s
analysis and proposal (Option 5)
|
Mark's
Comment
|
7
|
Closed
|
1.1 Spec
|
Interoperability
|
Design
|
Context-group "purpose"
recommended values
|
27
Jan 03 – Updated in 1.1 Draft 2d
10
Dec 02 : The proposal was unanimously approved.
Original Proposal:
Propose adding the following "purpose"
attribute values:
- location, The context-group is used to specify where the term was found
in the translatable source. Thus, it is not displayed.
- match, Specifies that the context information should be used during
translation memory lookups. Thus, it is not displayed.
- information, Specifies that the context is informational in nature,
indicating for example, how a term should be translated. Thus, should be
displayed to anyone editing the XLIFF.
Combinations of these values can be made via the standard mechanism of
XLIFF. Thus, purpose="location;match" would provide both location
and TM matching contextual information. The schemas for this are detailed
in the original suggestion URL
|
John Reid
|
John's
Proposal
|
|
8
|
Closed
|
1.1 Spec
|
Interoperability
|
Design
|
phase-name as
optional <alt-trans>
attribute
|
10
Dec 02: After the discussion on issue 4, this item was deemed to be
obsolete. Original Proposal: This was originally part of issue 4, but was
split out as its
own
issue on 3 Dec. meeting.
It
was observed by Yves that we had no
naming
convention for phase-name. Phase name
would
have at least 3 distict uses within alt trans:
1.
to identify a different suggestion from a TM,
2.
to capture the evolution of translation during the
translation
process, and
3.
identify rejected translations.
There
is no way of distinguishing these elements.
It
was agreed that we look at phase-name attribute in
relation
to this observation.
Proposal:
Add
the phase-name attribute to the alt-trans as an
optional
attribute. Following along with Yves's original
thoughts
on this, the phase-name could be placed at the
<alt-trans>
level for any <alt-trans> that has a <source>.
It
would be placed in the target of any <alt-trans> that
does
not have a <source>.
Example:
<trans-unit id="1" phase-name="5final">
<source>Cancel
Report</source>
<target
phase-name="4review">Annuler le Rapport</target>
<alt-trans
reason="Rejected-Inaccurate">
<target
phase-name="3trans">Annuler le rapport</target>
</alt-trans>
<alt-trans match-quality="50%"
reason="TM-Suggestion" phase-name="2pretrans">
<source>Cancel
All</target>
<target>Annuler
tout</target>
</alt-trans>
</trans-unit>
|
Yves / John Reid
|
Yves
observation
John
Reid's Proposal
|
|
9
|
Closed
|
1.1 Spec
|
Interoperability
|
Design
|
Attribute Enumerated Values
|
27
Jan 03 – Updated in 1.1 Draft 2d
7 Jan 2003 A motion from Doug was passed unanimously and
this stated all enumerated valued would be listed within the specification.
18 Dec 02: discussion is deadlocked
during XLIFF Teleconference. 17 Dec
02: Discussion on listing all enumerated values for attributes in the
specification (or not). At issue is whether these values are part of the
specification or part of an external schema.
|
Mark Levins
|
Mark's
proposal
|
|
10
|
Closed
|
1.1 Spec
|
Interoperability
|
Design
|
Whitespace / List item delimiters
|
27
Jan 03 – Updated in 1.1 Draft 2d
7
Jan 03
Mark’s proposal was agreed.
6 Jan 03:
Mark Levins submits revised text for the specification to be discussed at the
next teleconference:
D.2. Attribute Values
Attribute values are case sensitive. It is strongly recommended that
lower-case values are used. The specification recommends a number of values
for some attributes, these are all lower-case.
Where multiple attribute values are to be used in an XLIFF document, two
approaches are used:
For enumerated attributes (such as the 'purpose' attribute of
<context-group>) the separator must be a space.
For other textual attributes that are not validated, the specification
recommends the use of the semi-colon as a concatenation separator for
values, for example, multiple contacts may be listed for a <file>
with the attribute-value written thusly: contact-name="Frank
Sinatra;Sammy Davis Jnr;Dean Martin".
17 Dec 02:
Multiple attribute values (lists) and valid separators not certain if legal
to use ; or other delimiters, appears that whitespace is recommended by
W3C, but this does not
Preclude using or ,.
|
Mark Levins
|
Mark's
Initial Observation
Doug's
Suggestion
|
Mark’s
revised text for spec
|
11
|
Deferred
|
1.1 Spec
|
Interoperability
|
Design
|
TextContent Extensibility
|
14
Jan 03:
Gerard
moved for a vote on deferring TextContent Extensibility until the next revision
of XLIFF, Matt seconded and the motion was passed unanimously.
7
Jan 03
It was suggested at this meeting to defer discussion on this until after
Spec 1.1 is complete. This will be decided on the meeting at 14 Jan.
6 Jan 02: This issue was tabled after considerable discussion during
previous XLIFF teleconference. The discussion will continue at next
meeting (7 Jan). The main points from the discussion were:
1/XHTML can presently be handled within BPT and EPT inline tags
2/adding XHTML tags will create complexity and the requirement that all
XLIFF tools to be fully capable of interpreting XHTML tags.
17 Dec 02: Extensibility of the TextContent to allow non-XLIFF tags, for
example XHTML tags
|
Doug
|
Doug's
Proposal
|
Doug’s
Additional comments (17 Dec 02)
Doug’s
additional comments (24 Dec 02)
Yves’
Comments (24 Dec 02)
|
12
|
Closed
|
1.1 Spec
|
Specification Logistics
|
Specification Revision
|
Spec / Schema / DTD
|
27
Jan 03 – Updated in 1.1 Draft 2d
7 Jan 03 It was agreed that schema and specification changes
would be synchronized.
18 Dec 02: Spin-off from Issue 9. What is policy on
relationship between Specifications and Schemas / DTDs? Do minor changes to the DTD and Schema
require a revised specification, and visa versa? John Reid / Mark Levins
Issue raised during Issue 9 discussion at XLIFF
teleconference
|
John Reid / Mark Levins
|
Issue
raised during Issue 9 discussion at XLIFF teleconference
|
|
13
|
Closed
|
1.1 Spec
|
Interoperability
|
Design
|
Add phase-name attribute
to <count>
|
27
Jan 03 – Updated in 1.1 Draft 2d
#7
Jan 03
This was agreed
Proposal: Add phase-name attribute to <count>
We already have a <phase> element that stores the tool-id used in
that phase. The phase-name attribute could be added to <count>. Thus,
when that count was produced and by what, could be ascertained by any subsequent
tool and a determination of it to use the count could be made.
|
John Reid
|
John’s
Proposal
|
|
14
|
Open Unassigned
|
1.1 Spec
|
Deliverables
|
Schema
|
Validation of XLIFF 1.0
with 1.1 Schema
|
11
Feb 2003 Discuss and possibly vote on this issue.
Keep
the targetNamespace in the schema (as it
is
now), and document that XLIFF 1.0 documents will need to specify the
XLIFF1.1 namespace in order to validate.
|
Doug
|
Doug's
proposal (31 Jan 03)
|
OASIS
Namespace Guideline
More
examples
|
15
|
Open
Unassigned
|
1.1 Spec
|
Interoperability
|
Design
|
Bin Unit size
|
Where
we have icons that are being localized, we need to keep track of the change
in size of the icon. However, the current definition of <bin-unit>
does not have that information. In the <trans-unit> we have the coord
attribute to store that info. Since we are at a 1.0 implementation, we
could use a <prop> to give us that info but I prefer the
<context> since we can codify it and make it understandable to
translators, provided that the tools understand it.
<bin-unit id="1"
mime-type="image/gif" translate="yes"
reformat="yes">
<bin-source><external-file
href=""btnadvanced.gif"/></bin-source>
<context-group name="translation">
<context
context-type="bin-coord">0;0;86;16</context>
</context-group>
</bin-unit>
The
better solution would be to add coord as an attribute of <bin-unit>
and <bin-target>. After all, what
does reformat control for <bin-unit>? The only attributes that
could be reformatted are mime-type, restype, and resname. We may need the
other following attributes:
coord <bin-unit> and
<bin-target>
size-unit <bin-unit> only
maxheight <bin-unit> only
minheight <bin-unit> only
maxwidth <bin-unit> only
minwidth <bin-unit> only
The
above could then be expressed as follows.
<bin-unit id="1"
mime-type="image/gif" translate="yes"
reformat="yes" coord="0;0;86;16">
<bin-source><external-file
href=""btnadvanced.gif"/></bin-source>
</bin-unit>
If
someone has come across this same problem and found a different solution, I
would appreciate the help. Otherwise, as much as I don't want to delay the
process, we may want to consider taking care of this now. What do you
think?
|
John Reid
|
John's
original proposal
|
Yves’
Follow Up
Doug’s
input
Yve’s
next follow up
Stephen
Holmes’ input
|
16
|
Open Unassigned
|
1.1 Spec
|
Interoperability
|
Design
|
Attribute name Case guidelines
|
11
Feb 2003 Discuss and vote
Removing
the guidelines about using lowercase values for attributes. The first
paragraph of section D.2. As it
is
part of the 1.0 specification we need vote to accept it.
|
Yves & Christian
|
Yves’
original mail
|
|
17
|
Open Unassigned
|
1.1 Spec
|
Deliverables
|
Schema
|
Extension Namespace
|
One little (but important) point about the extension
points in the schema: They are currently set as '##other', which means
"Any well-formed XML that is not from the target namespace of the type
being defined" (thus, non XLIFF). Christian reminded me that a while
back, there was a discussion
about replacing this by '##any', which means "Any
well-formed XML from any namespace" (thus, including XLIFF itself).
The rational was that we may want give users a possiblity to use XLIFF data
categories in a non-standard context. See the W3C documents for more
information on namespaces at http://www.w3.org/TR/xmlschema-0/#nsTable.
Input on this would be welcome
|
Yves
|
Yves’
original mail
|
|
18
|
Open Unassigned
|
1.1 Spec
|
Interoperability
|
Design
|
context-type
|
As per yesterday discussion it seems that some of us
assume the 'context-type' attribute can take several values (like
'purpose'). There is no information about this is the current
specifications and the current
schema doesn't allow this. If it's the case, the schema
needs to be adapted and use a pattern like for 'purpose'. When reading the
specifications keep
in mind that if the description of the attribute does
not mention the values can be used in combination it means the schema does
not allow combination.
Let me know where else there is a discripancy like for
'context-type'.
|
Yves
|
Yves’
original mail
|
|
19
|
Open Unassigned
|
1.1 Spec
|
Interoperability
|
Design
|
Consolidated Attribute
Values Lists
|
The current proposals we have been discussing for new
values for this list of attributes raise the issues of cross-pollination of
attribute values from one attribute into the others. As it currently
stands, a number of attributes values are found in other categories for a
number of reasons.
Solution:
1. Add 3 attributes values to the count-type
attribute: datatype, restype, and state. Each attribute value must augment
its value by adding a dot and an enumerated value from their corresponding
attribute. Here is an example: count-type=datatype.cstring, or
count-type=restype.dialog, or count-type=state.needs-translation.
2. Remove duplicated attribute values in count-type:
a.cstring, msglib (in datatype)
b.needs-translation, needs-review, signed-off (in
state). Not sure about Christian's
new proposed list. They appear to be count-type in their own right.
Comment?
3. Datatype, restype, and state attributes have
additional values proposed, which won't discussed here.
|
Gerard
|
Gerard’s
Original Proposal
|
|
20
|
Open Unassigned
|
1.1 Spec
|
Interoperabilty
|
Design
|
Phases of Alt Trans
|
Yves originally suggested using the existence of the
match-quality attribute to determine whether an alt-trans was leveraged or
change control. Match-quality is free text that can contain a score or any
arbitrary value based on the tool that generates the <alt-trans>.
Unfortunately this makes it difficult to rely on that attribute.
Mark suggested adding a reason attribute to the
<alt-trans> with the values 'TM Suggestion', 'MT Suggestion',
'Rejected-Inaccurate', 'Rejected-Spelling', 'Rejected-Grammar' and
'Rejected-Length'. This would allow us to mark an alt-trans as being
leveraged ('TM Suggestion' and 'MT Suggestion') or change-control
('Rejected-Inaccurate', 'Rejected-Spelling', 'Rejected-Grammar' and
'Rejected-Length'). Thus, all <target>s in an <alt-trans> would
have to have the same reason. Or, rather, a new <alt-trans> would be
needed for each of the rejected reasons. This may create a lot of
<alt-trans> but only if someone (the translator?) is doing a poor
job. Likely there will only be very few of these.
I had suggested we use the origin attribute of
<alt-trans> with the value 'this-file' to indicate a change-control.
This would require enumerating that one value for origin and making any
other values to begin with an 'x-', to be consistent. That just isn't very
practical.
Maybe a combination Yves's and Mark's suggestions are
the answer. Enumerate match-quality with Mark's values and allow extension.
I still stand by my suggestion of adding a state
attribute to <trans-unit> for the reasons outlined. I also suggested
to add Mark's reason values to state. However, I suggest that we not do
that.
|
John Reid
|
John’s
original proposal
|
Doug’s
Comment
John’s
reply to Doug
|
21
|
Open Unassigned
|
1.1 Spec
|
Interoperabilty
|
Design
|
Attribute Values List
|
Define the list of list of attribute values.
|
|
Tony’s
consolidated list
|
|