Home » Modeling » TMF (Xtext) » [Xtext] Can any compare Xtext with TCS?
| [Xtext] Can any compare Xtext with TCS? [message #43896] |
Sun, 17 May 2009 12:48  |
Eclipse User |
|
|
|
Hi,
It seems that XText is quite similar to TCS in terms of functionality. If somebody can compare them two, I think it is better for newbies to have an overview of them.
Thanks.
Michael.
|
|
|
| Re: [Xtext] Can any compare Xtext with TCS? [message #43992 is a reply to message #43896] |
Mon, 18 May 2009 05:36  |
| Eclipse User |
|
|
|
Michael,
I've used both TCS and Xtext, although I'm more familiar with Xtext.
First
the technical differences.
I find TCS more complete for defining a closed DSL; that is one that
does
not have in-built primitives or use import clauses. It also within the
grammar
of the language definition allows the definition of the language objects
that
create scope and the language objects that are added to the current scope.
This means that it is possible to create many DSLs with the need to write
any code. However, if the generated implementation of a DSL does not
perform as you need, it is not easy to modify.
Xtext is probably better for non-closed DSLs; i.e. ones that have
primitive
types and use import clauses. It is also has mechanisms designed to permit
the customisation the generated implementation; in my view, the use of
dependency injection and adapters means that it is not always obvious as
to how the customisation code has to interact with the generated code. As I
see it, the main drawback of Xtext is that it assumes that a resource is a
flat
namespace. Hence, if a DSL has any scope concepts, it is necessary to
write customisation code.
Now for the non-technical comparison. The activity in this news group
appears to show that TCS is no longer under active development or support.
The source, for example, has not been accessible in an Eclipse repository
for almost a year. Hence, for any new project, I would recommend that you
only consider Xtext.
Andy.
"Michael" <my25@macs.hw.ac.uk> wrote in message
news:gupf5g$ul0$1@build.eclipse.org...
> Hi,
> It seems that XText is quite similar to TCS in terms of functionality.
> If somebody can compare them two, I think it is better for newbies to have
> an overview of them.
> Thanks.
> Michael.
|
|
|
| Re: [Xtext] Can any compare Xtext with TCS? [message #44023 is a reply to message #43992] |
Mon, 18 May 2009 08:06 |
| Eclipse User |
|
|
|
Andy Carpenter wrote:
> Michael,
>
> I've used both TCS and Xtext, although I'm more familiar with Xtext.
> First
> the technical differences.
>
> I find TCS more complete for defining a closed DSL; that is one that
> does
> not have in-built primitives or use import clauses. It also within the
> grammar
> of the language definition allows the definition of the language objects
> that
> create scope and the language objects that are added to the current scope.
> This means that it is possible to create many DSLs with the need to write
> any code. However, if the generated implementation of a DSL does not
> perform as you need, it is not easy to modify.
>
> Xtext is probably better for non-closed DSLs; i.e. ones that have
> primitive
> types and use import clauses. It is also has mechanisms designed to permit
> the customisation the generated implementation; in my view, the use of
> dependency injection and adapters means that it is not always obvious as
> to how the customisation code has to interact with the generated code. As I
> see it, the main drawback of Xtext is that it assumes that a resource is a
> flat
> namespace. Hence, if a DSL has any scope concepts, it is necessary to
> write customisation code.
>
> Now for the non-technical comparison. The activity in this news group
> appears to show that TCS is no longer under active development or support.
> The source, for example, has not been accessible in an Eclipse repository
> for almost a year. Hence, for any new project, I would recommend that you
> only consider Xtext.
>
> Andy.
>
>
> "Michael" <my25@macs.hw.ac.uk> wrote in message
> news:gupf5g$ul0$1@build.eclipse.org...
>> Hi,
>> It seems that XText is quite similar to TCS in terms of functionality.
>> If somebody can compare them two, I think it is better for newbies to have
>> an overview of them.
>> Thanks.
>> Michael.
>
>
Hi Andy:
Thanks, it is very useful for me. I tried to make TCS work in my Eclipse IDE for about one week but still failed.
I think it is wise for me to turn to Xtext. According to the documentation of XText, it has a kind of metamodel inference mechanism
which generates ecore metamodel based on grammar definition. However, I have already composed the ecore metamodel of my DSL,
hence I do not want to adopt the metamodel inference mechanism. Could Xtext handle this situation well?
Thanks again.
Michael.
|
|
|
| Re: [Xtext] Can any compare Xtext with TCS? [message #44054 is a reply to message #44023] |
Mon, 18 May 2009 09:24 |
| Eclipse User |
|
|
|
Hi Michael,
Xtext is perfectly ok with existing Metamodels. Please have a look at
the reference documentation at
http://wiki.eclipse.org/Xtext/Documentation#EPackage_import
However, please make sure, that your EPackage is valid according to the
EMF validation.
If you have any further questions about Xtext, feel free to ask them in
the newsgroup. We'll be glad to help you with any problem that occurs.
Btw: If may want to download the distro from
http://oaw.itemis.com/openarchitectureware/language=en/2837/ downloads as
a jump start.
Regards,
Sebastian
Am 18.05.2009 14:06 Uhr, schrieb Michael:
> Andy Carpenter wrote:
>> Michael,
>>
>> I've used both TCS and Xtext, although I'm more familiar with Xtext.
>> First
>> the technical differences.
>>
>> I find TCS more complete for defining a closed DSL; that is one that does
>> not have in-built primitives or use import clauses. It also within the
>> grammar
>> of the language definition allows the definition of the language
>> objects that
>> create scope and the language objects that are added to the current
>> scope.
>> This means that it is possible to create many DSLs with the need to write
>> any code. However, if the generated implementation of a DSL does not
>> perform as you need, it is not easy to modify.
>>
>> Xtext is probably better for non-closed DSLs; i.e. ones that have
>> primitive
>> types and use import clauses. It is also has mechanisms designed to
>> permit
>> the customisation the generated implementation; in my view, the use of
>> dependency injection and adapters means that it is not always obvious as
>> to how the customisation code has to interact with the generated code.
>> As I
>> see it, the main drawback of Xtext is that it assumes that a resource
>> is a flat
>> namespace. Hence, if a DSL has any scope concepts, it is necessary to
>> write customisation code.
>>
>> Now for the non-technical comparison. The activity in this news group
>> appears to show that TCS is no longer under active development or
>> support.
>> The source, for example, has not been accessible in an Eclipse repository
>> for almost a year. Hence, for any new project, I would recommend that you
>> only consider Xtext.
>>
>> Andy.
>>
>>
>> "Michael" <my25@macs.hw.ac.uk> wrote in message
>> news:gupf5g$ul0$1@build.eclipse.org...
>>> Hi,
>>> It seems that XText is quite similar to TCS in terms of
>>> functionality. If somebody can compare them two, I think it is better
>>> for newbies to have an overview of them.
>>> Thanks.
>>> Michael.
>>
>>
> Hi Andy:
> Thanks, it is very useful for me. I tried to make TCS work in my Eclipse
> IDE for about one week but still failed.
> I think it is wise for me to turn to Xtext. According to the
> documentation of XText, it has a kind of metamodel inference mechanism
> which generates ecore metamodel based on grammar definition. However, I
> have already composed the ecore metamodel of my DSL,
> hence I do not want to adopt the metamodel inference mechanism. Could
> Xtext handle this situation well?
> Thanks again.
> Michael.
|
|
|
| Re: [Xtext] Can any compare Xtext with TCS? [message #44215 is a reply to message #44054] |
Mon, 18 May 2009 13:57 |
| Eclipse User |
|
|
|
Hi, Sebastian:
I am still in the stage of teh preliminary study of Xtert. I would like to incorporate a model injector and extractor into a standalone ATL tranformation application.
So my question is whether Xtext can generate them based on the grammar specification of the concerned DSL. If so, are they handy to be incorporated in the standalone
ATL transformation application?
You help is highly appreciated.
Regards.
Michael.
> Hi Michael,
>
> Xtext is perfectly ok with existing Metamodels. Please have a look at
> the reference documentation at
> http://wiki.eclipse.org/Xtext/Documentation#EPackage_import
>
> However, please make sure, that your EPackage is valid according to the
> EMF validation.
>
> If you have any further questions about Xtext, feel free to ask them in
> the newsgroup. We'll be glad to help you with any problem that occurs.
>
> Btw: If may want to download the distro from
> http://oaw.itemis.com/openarchitectureware/language=en/2837/ downloads as
> a jump start.
>
> Regards,
> Sebastian
>
>
>
> Am 18.05.2009 14:06 Uhr, schrieb Michael:
>> Andy Carpenter wrote:
>>> Michael,
>>>
>>> I've used both TCS and Xtext, although I'm more familiar with Xtext.
>>> First
>>> the technical differences.
>>>
>>> I find TCS more complete for defining a closed DSL; that is one that
>>> does
>>> not have in-built primitives or use import clauses. It also within the
>>> grammar
>>> of the language definition allows the definition of the language
>>> objects that
>>> create scope and the language objects that are added to the current
>>> scope.
>>> This means that it is possible to create many DSLs with the need to
>>> write
>>> any code. However, if the generated implementation of a DSL does not
>>> perform as you need, it is not easy to modify.
>>>
>>> Xtext is probably better for non-closed DSLs; i.e. ones that have
>>> primitive
>>> types and use import clauses. It is also has mechanisms designed to
>>> permit
>>> the customisation the generated implementation; in my view, the use of
>>> dependency injection and adapters means that it is not always obvious as
>>> to how the customisation code has to interact with the generated code.
>>> As I
>>> see it, the main drawback of Xtext is that it assumes that a resource
>>> is a flat
>>> namespace. Hence, if a DSL has any scope concepts, it is necessary to
>>> write customisation code.
>>>
>>> Now for the non-technical comparison. The activity in this news group
>>> appears to show that TCS is no longer under active development or
>>> support.
>>> The source, for example, has not been accessible in an Eclipse
>>> repository
>>> for almost a year. Hence, for any new project, I would recommend that
>>> you
>>> only consider Xtext.
>>>
>>> Andy.
>>>
>>>
>>> "Michael" <my25@macs.hw.ac.uk> wrote in message
>>> news:gupf5g$ul0$1@build.eclipse.org...
>>>> Hi,
>>>> It seems that XText is quite similar to TCS in terms of
>>>> functionality. If somebody can compare them two, I think it is better
>>>> for newbies to have an overview of them.
>>>> Thanks.
>>>> Michael.
>>>
>>>
>> Hi Andy:
>> Thanks, it is very useful for me. I tried to make TCS work in my Eclipse
>> IDE for about one week but still failed.
>> I think it is wise for me to turn to Xtext. According to the
>> documentation of XText, it has a kind of metamodel inference mechanism
>> which generates ecore metamodel based on grammar definition. However, I
>> have already composed the ecore metamodel of my DSL,
>> hence I do not want to adopt the metamodel inference mechanism. Could
>> Xtext handle this situation well?
>> Thanks again.
>> Michael.
>
|
|
|
| Re: [Xtext] Can any compare Xtext with TCS? [message #44245 is a reply to message #44215] |
Mon, 18 May 2009 14:28 |
| Eclipse User |
|
|
|
Hi Michael,
I (and I guess the same is true for Sebastian) don't know the
ATL-specific concepts, but as far as I could find out in ATL (TCS)
an injector is a parser and
an extractor is a generator/pretty printer/serializer.
If that is coorect, I guess the answer is 'yes'. As long as you have a
valid complete Xtext grammar of your language.
I also don't know whether it is convenient to use them in the context of
ATL (but would be interested top learn), but if ATL works with EMF
Resources (I guess so) it should be straight forward.
Xtext provides you with an implementation of an EMF Resource, where the
load uses the parser and the save uses the serializer.
For the serializer (unlike TCS) the pretty-printing information is
declared outside the grammar (documentation for that aspect will be
available soon).
Hope that helps,
Sven
Michael schrieb:
> Hi, Sebastian:
> I am still in the stage of teh preliminary study of Xtert. I would
> like to incorporate a model injector and extractor into a standalone
> ATL tranformation application.
> So my question is whether Xtext can generate them based on the grammar
> specification of the concerned DSL. If so, are they handy to be
> incorporated in the standalone
> ATL transformation application?
> You help is highly appreciated.
> Regards.
> Michael.
>> Hi Michael,
>>
>> Xtext is perfectly ok with existing Metamodels. Please have a look at
>> the reference documentation at
>> http://wiki.eclipse.org/Xtext/Documentation#EPackage_import
>>
>> However, please make sure, that your EPackage is valid according to
>> the EMF validation.
>>
>> If you have any further questions about Xtext, feel free to ask them
>> in the newsgroup. We'll be glad to help you with any problem that occurs.
>>
>> Btw: If may want to download the distro from
>> http://oaw.itemis.com/openarchitectureware/language=en/2837/ downloads
>> as a jump start.
>>
>> Regards,
>> Sebastian
>>
>>
>>
>> Am 18.05.2009 14:06 Uhr, schrieb Michael:
>>> Andy Carpenter wrote:
>>>> Michael,
>>>>
>>>> I've used both TCS and Xtext, although I'm more familiar with Xtext.
>>>> First
>>>> the technical differences.
>>>>
>>>> I find TCS more complete for defining a closed DSL; that is one that
>>>> does
>>>> not have in-built primitives or use import clauses. It also within the
>>>> grammar
>>>> of the language definition allows the definition of the language
>>>> objects that
>>>> create scope and the language objects that are added to the current
>>>> scope.
>>>> This means that it is possible to create many DSLs with the need to
>>>> write
>>>> any code. However, if the generated implementation of a DSL does not
>>>> perform as you need, it is not easy to modify.
>>>>
>>>> Xtext is probably better for non-closed DSLs; i.e. ones that have
>>>> primitive
>>>> types and use import clauses. It is also has mechanisms designed to
>>>> permit
>>>> the customisation the generated implementation; in my view, the use of
>>>> dependency injection and adapters means that it is not always
>>>> obvious as
>>>> to how the customisation code has to interact with the generated code.
>>>> As I
>>>> see it, the main drawback of Xtext is that it assumes that a resource
>>>> is a flat
>>>> namespace. Hence, if a DSL has any scope concepts, it is necessary to
>>>> write customisation code.
>>>>
>>>> Now for the non-technical comparison. The activity in this news group
>>>> appears to show that TCS is no longer under active development or
>>>> support.
>>>> The source, for example, has not been accessible in an Eclipse
>>>> repository
>>>> for almost a year. Hence, for any new project, I would recommend
>>>> that you
>>>> only consider Xtext.
>>>>
>>>> Andy.
>>>>
>>>>
>>>> "Michael" <my25@macs.hw.ac.uk> wrote in message
>>>> news:gupf5g$ul0$1@build.eclipse.org...
>>>>> Hi,
>>>>> It seems that XText is quite similar to TCS in terms of
>>>>> functionality. If somebody can compare them two, I think it is better
>>>>> for newbies to have an overview of them.
>>>>> Thanks.
>>>>> Michael.
>>>>
>>>>
>>> Hi Andy:
>>> Thanks, it is very useful for me. I tried to make TCS work in my Eclipse
>>> IDE for about one week but still failed.
>>> I think it is wise for me to turn to Xtext. According to the
>>> documentation of XText, it has a kind of metamodel inference mechanism
>>> which generates ecore metamodel based on grammar definition. However, I
>>> have already composed the ecore metamodel of my DSL,
>>> hence I do not want to adopt the metamodel inference mechanism. Could
>>> Xtext handle this situation well?
>>> Thanks again.
>>> Michael.
>>
|
|
| |
| Re: [Xtext] Can any compare Xtext with TCS? [message #44853 is a reply to message #44719] |
Fri, 22 May 2009 13:44 |
| Eclipse User |
|
|
|
Formatting is one of the few things where we'll have late changes /
fixes. Expect working formatting and correpsonding documentation by RC3
(two weeks).
Cheers,
Sven
Andy Carpenter schrieb:
> Snip
>
>> For the serializer (unlike TCS) the pretty-printing information is
>> declared outside the grammar (documentation for that aspect will be
>> available soon).
>
> Is any draft documentation on controlling the formatting
> available, or hints on where it is configured? At the moment
> I'm getting all of my text on a single line.
>
> Andy.
>
>
|
|
|
Goto Forum:
Current Time: Mon Jul 14 14:45:24 EDT 2025
Powered by FUDForum. Page generated in 0.09564 seconds |
Home