chicken-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Chicken-setup redesign (was: Re: [Chicken-users] Re: getopt, getopt_


From: minh thu
Subject: Re: Chicken-setup redesign (was: Re: [Chicken-users] Re: getopt, getopt_long?)
Date: Mon, 14 Jul 2008 14:53:28 +0200

2008/7/14 felix winkelmann <address@hidden>:
> On Mon, Jul 14, 2008 at 2:35 PM, Mario Domenech Goulart
> <address@hidden> wrote:
>>>
>>> Yes, that would work. You would also need a file for foo-1, I suppose,
>>> if one wants to to load the highest minor version of 1.XXX. This will
>>> produce lots of little loader-files, though.
>>
>> Hmmm.  I haven't thought about this level of granularity.  I thought
>> users would either specify the exact version they want or no version
>> (in this case, the highest would be used).  How would users specify
>> they want to load the highest minor version of 1.XXXX?
>>
>> I was thinking about creating a little loader for the highest version
>> _only_, for when users don't specify the version when loading an
>> extension (so the highest is picked).
>>
>> For example, if the foo egg is installed (say its version is 1.0), the
>> local repo would contain:
>>
>>    foo-1.0.so
>>    foo.so
>>
>> foo.so would be generated by compiling
>>
>>   (require-library foo-1.0)
>>
>> since 1.0 is the highest version.
>>
>> Now suppose there's a new version for foo (1.1).  So, when foo is
>> updated, the local repo would contain:
>>
>>    foo-1.0.so
>>    foo-1.1.so
>>    foo.so
>>
>> Now, foo.so would contain the code to load foo-1.1, not 1.0, but users
>> will still be able to load 1.0 if they need (explicitly, though).
>>
>
> I guess having eitther the newest or a specific version is enough. I don't
> think having finer grained access is really necessary, but I think Peter
> proposed that?

I think he did too. But is you're view of fine-grained a bit 'reversed' ?

If you say you want foo-1, and *if* that library is done such that minor
version increases are backward-compatible in the same major number,
isn't it more naturall to request foo-1 and then get the highest minor
number than requesting foo then have the higher overall number ?

(since higher major numbers are more likely to break compatibility
than higher minor numbers)

In other words, foo-1.1 and foo-1.2 are the same library, while
foo-1.0 and foo-2.0 are too different. If I make a mistake here, I guess
we could have some guidelines on the semantic of version numbers...

Cheers,
thu




reply via email to

[Prev in Thread] Current Thread [Next in Thread]