[30474] trunk/dports/python/py-reportlab/Portfile

[Ryan Schmidt]

On Oct 29, 2007, at 08:20, Weissmann Markus wrote:

> On 29.10.2007, at 05:51, Ryan Schmidt wrote:
>
>> On Oct 28, 2007, at 17:43, source_changes at macosforge.org wrote:
>>
>> [snip]
>>
>>> -distfiles		ReportLab_1_20.tgz
>>> -checksums		md5 8d374b9b6f7db3fc0821082347d5e6b5
>>> +distfiles		ReportLab_2_1.tgz
>>> +checksums		sha1 67c6fba57b6372987dd3280be257a1f6bb7fff80
>>> +depends_lib-append	port:py-numeric
>>>
>>> -worksrcdir		reportlab_1_20/reportlab
>>> +worksrcdir		reportlab_2_1/reportlab
>>
>> [snip]
>>
>> Instead of hardcoding the version in py-reportlab and py25- 
>> reportlab, you may want
>>
>> distfiles ReportLab_[strsed ${version} {g/[.]/_/}].tgz
>>
>> and
>>
>> worksrcdir reportlab_[strsed ${version} {g/[.]/_/}]/reportlab
>
> Yes, except that this is very hard to read while the benefit of not  
> having to change a 2nd line for an update is nil in my eyes.

Ok.


>> Also, when you created the py25-reportlab portfile based on the py- 
>> reportlab portfile, it would have been better to "svn cp py- 
>> reportlab py25-reportlab" in the python directory, rather than  
>> copy it outside of Subversion and then add it. Had you done that,  
>> "svn log `port file py25-reportlab`" would now be more useful  
>> (would reflect the port's history back thru the creation of the  
>> original py-reportlab portfile in 2003) instead of appearing to  
>> have been called into existence just now with no history.
>
> In general you may be right -- but what did you read from the log  
> of py-reportlab that could be interesting for py25-reportlab?

Nothing. I'm not interested in Python at this time. But with other  
portfiles, I often look through the history with svn log and look at  
the output of svn blame to determine why something is in the  
portfile. By adding the file as a new file, rather than copying it,  
you prevent anyone from learning anything using these means.

In Subversion's eyes, it looks like you have created a completely new  
portfile. But you haven't. You based it on a previous and related  
portfile. That should be reflected in the history by using svn cp.

In this case it doesn't matter since portfiles are so small, but  
using svn cp also saves disk space in the repository, since it only  
needs to store a diff to the previous version instead of storing the  
complete text of the file again.