pkgsrc-Users archive

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

Re: Anti-bundling materials



On 8/21/21 8:45 PM, J. Lewis Muir wrote:
On 08/21, Jason Bacon wrote:
On 8/21/21 12:33 PM, J. Lewis Muir wrote:
As you said, this practice "leads to security issues and other bugs that
are difficult to fix because the software uses an outdated API."  I'm
not sure what you mean by "difficult to fix because the software uses an
outdated API" (I would think that would actually make the software more
likely to keep working because it has bundled the library with the API
that it uses), but in general, the exact same issues, and more outlined
in some of the resources you posted upthread, exist for the 99% that are
not being addressed.

What I'm saying here is the bundled library *is* the problem since it has
known vulnerabilities or bugs, and we can't just hack the build system to
use an up-to-date replacement from pkgsrc since the API is different.  There
is one such tool I won't name that bundles an outdated SSL library, which
people are using to process private health information.

Got it.  Then to me that's just an example of one of the reasons
why it's bad to bundle.  IMO, when a project bundles, they take on
responsibility for everything they bundle.  It's their responsibility to
monitor the software they bundle for security vulnerabilities, update
what they bundle if there is a security vulnerability, and make a new
release.  That's not your problem; that's the project that bundles'
problem.

As far as I'm concerned, the security vulnerability should be reported
against the project that bundles, and they should fix it (by patching or
updating the version they bundle).  If they don't fix it, then they'll
get a reputation for not caring about security, and users can choose
whether they're OK with that.

Lewis

I agree with all of this, except the statement that it's not my problem. Any time I encounter a bundled library, I feel obligated to explore the option of unbundling it so we end up with a higher quality package. So at minimum, it's taking a bite out of my valuable time.

Hopefully seeing a broad consensus against this practice among package managers will diminish its use. Most of the developers I've encountered who do this have no idea about the risks, so a little education might be all it takes to sway them.



Home | Main Index | Thread Index | Old Index