[NTLUG:Discuss] apt-get, rpm and library versioning (was: Debian)
Leroy Tennison
leroy_tennison at prodigy.net
Sat Jun 25 01:10:53 CDT 2005
Burton Strauss wrote:
>
>
>-----Original Message-----
>From: discuss-bounces at ntlug.org [mailto:discuss-bounces at ntlug.org] On Behalf
>Of Leroy Tennison
>Sent: Thursday, June 23, 2005 11:09 PM
>To: NTLUG Discussion List
>Subject: Re: [NTLUG:Discuss] apt-get, rpm and library versioning (was:
>Debian)
>
>Burton Strauss wrote:
>
><snip />
>
>
>
>>You know, I think it's time to stop trying to simplify our life. If
>>
>>
>SuperDuperFunction is going to be
>
>
>>so much different than SuperFunction then create >another library. Yes,
>>
>>
>this profilerates 'executables'
>
>
>>but allows older code to still work without creating a 'jet-powered
>>
>>
>stagecoach' combination of the latest
>
>
>>and greatest whiz-bang features along with backwards compatibility. DASD
>>
>>
>is getting really cheap, may
>
>
>>as well use it.
>>
>>
>
>
>Nope, that's no better.
>
>You end up with 100s of versions of popular libraries hard linked into
>programs (that's what STATIC linking does).
>
>Then when you find a problem in, say, OpenSSL, you have to replace EVERY
>single hard-linked package. Miss ONE and you are exposed to the security
>problems. Dynamic linking means you only have to replace the one .so file.
>
>
>-----Burton
>
>
>
>_______________________________________________
>https://ntlug.org/mailman/listinfo/discuss
>
>
>
I realize that, at some point, this is a "would you rather be shot or
hung" kind of issue. However, if we had a *** few *** more libraries it
might make things easier. Those programs which needed a specific
version would refer to it (or provide an override to do so), those
programs which didn't care could use a "versionless" reference.
Granted, "versionless" could become version dependent at any time which
is why the override feature should probably be built in from the start.
This approach would be a balancing act and I realize it would never be
perfect but it might get us closer to a "least worst" situation.
More information about the Discuss
mailing list