[flexi-streams-devel] a patch for chineses's cp936(gbk) encoding support.

[Edi Weitz]

Thanks, I'll take a look at this when I'm less busy than now.

Again, please use the mailing list for patches and questions - see Cc.

Thanks,
Edi.


On Fri, Feb 5, 2010 at 11:31 AM, jingtaozf <jingtaozf at gmail.com> wrote:
> Hi Edi;
>   I have modifed the code to fit your request.and the encode name changed from cp936 to gbk.
>   The following is the response.
>
> 1. There are several new functions and macros which don't have a
> documentation string.
> ==> i have add document to function get-multibyte-mapper,others are referenced with your codes.
> 2. There are tabs in the files and sometimes the indentation seems wrong.
> ==> i have removed tabs.
> 3. There are parts of the code which are commented out.  If they
> aren't used, they shouldn't be in there.
> ==> yes,i deleted them.
> 4. There's at least one file (seems like a variant of the ASDF system
> definition) which shouldn't be in there at all.
> ==>sorry,but i don't know what you mean?
> 5. The link in the HTML documentation is wrong.
> ==>i fixed the url link to gbk.
> 6. Isn't get-multibyte-mapper the wrong name?  You don't get the mapper, do you?
> ==>yes,this function is borrowed from sbcl's source code,and i have pointed out this in the comment.
> 7. And the function itself looks like Scheme to me.  I think it'd be
> easier to understand using normal iteration.
> ==>i have re-write the decode and encode part.but i am not familiar with scheme. :(
> 8. I don't understand why the encoding factor is 1.5.  Is the comment
> correct or just copied?
> ==>i don't know how to set the encoding factor,gbk(cp936) has one or two octets.i set it to 1.5 with my estimate.
> 9. I'm not familiar with CP936.  Is it correct that there's only a
> big-endian version?
> ==>gbk(cp936) has no endian problem, i borrowed code from your utf16's implemention,now i have fixed it.
> 10. One of the new files contains DOS line endings.
> ==> some your source files has DOS line endings(such as ascii.lisp),some use unix line endings.
>      i don't know why and maybe i modified it un-expectly.
>
>
> Fothermore,this patch file is encoded use utf-8,because my test file contains file which is encoded by gbk format.
> I also put the test files(with gbk format) to the attatchment,because i am not sure the patch will generate them correctly.
>
> 2010-02-04
>
>
>
> jingtaozf
>
>
>
> 发件人： Edi Weitz
> 发送时间： 2009-10-29  23:08:38
> 收件人： jingtao xu
> 抄送： General interest list about flexi-streams
> 主题： Re: a patch for chineses's cp936(gbk) encoding support.
>
> On Thu, Oct 29, 2009 at 3:17 PM, jingtao xu <jingtaozf at gmail.com> wrote:
>> hi,Edi Weitz:
>>    Because i want to use drakma as my web development tool,but it use
>> flexi-streams and could not decode chinese's cp936 characters,so i make  a
>> patch to flexi-streams.
>>
>>    The sbcl has supported cp936(by files supported by beinghe,which cvs path
>> is:sbcl/src/code/external-formats/enc-cn.lisp,enc-cn-tbl.lisp). i use file
>> enc-cn-tbl.lisp here.
>>
>>   The patches is make by following commands:
>>      diff -urN ~/.sbcl/site/flexi-streams-1.0.7/ ~/Downloads/flexi-streams/
>>>flexi-streams-cp936.patch
>>
>>   I send you both the patch files and the patched codes.
>>  The test codes is added and passed.
>>  The documention is updated.
> [I've put the mailing list on Cc where I think we should continue this.]
> First of all, I'd be happy to include this into the flexi-streams
> distribution.  There are a couple of issues with the patch, though.
> What I saw from briefly looking at it:
> 1. There are several new functions and macros which don't have a
> documentation string.
> 2. There are tabs in the files and sometimes the indentation seems wrong.
> 3. There are parts of the code which are commented out.  If they
> aren't used, they shouldn't be in there.
> 4. There's at least one file (seems like a variant of the ASDF system
> definition) which shouldn't be in there at all.
> 5. The link in the HTML documentation is wrong.
> 6. Isn't get-multibyte-mapper the wrong name?  You don't get the mapper, do you?
> 7. And the function itself looks like Scheme to me.  I think it'd be
> easier to understand using normal iteration.
> 8. I don't understand why the encoding factor is 1.5.  Is the comment
> correct or just copied?
> 9. I'm not familiar with CP936.  Is it correct that there's only a
> big-endian version?
> 10. One of the new files contains DOS line endings.
> There's probably more.
> If you could go over these and send a cleaned-up version of the patch,
> I'll review it.  FWIW, here are some guidelines:
>  http://weitz.de/patches.html
> Thanks,
> Edi.
>