mschwendt at gmail.com
Sat Jul 11 22:09:07 UTC 2015
On Sat, 11 Jul 2015 11:45:13 -0600, Stephen John Smoogen wrote:
> I agree that lowering hurdles is not good and I don't want to lower
> hurdles, I want to fix broken ones.
> 1) Parts of reviews come across as arbitrary nitpicks. You put a
> package into 3 different reviewers.. you will get 3 different sets of
> fixes that seem to need to be made.
And if you re-review packages, which are included in the distribution
already, you run into bad examples. It happens occasionally that new
packagers copy from an existing Fedora spec file in good hope that it
would meet the packaging guidelines, but it doesn't.
For a more serious response: I don't know how to fix the reviewers, who
still request packagers to apply special indentation or other cosmetic
changes to a spec file. However, it is good that some reviewers are
experienced and point out pitfalls and mistakes not covered by the
guideslines [yet]. These are cases that may develop into somebody
asking the FPC to extend/enhance/improve the guidelines.
> 2) Reviews and fixes against existing packages aren't done. So if I
> base my package off of various high profile packages in Fedora already
> I won't pass review..
That's sounds like what I've pointed out above. Some people like
convoluted spec files and "strange" packaging, and there's not much
ordinary packagers can do about that. Some people love argueing about
the guidelines, such as where to place .so symlinks. Some love coming
up with reasons to violate the guidelines. It's the FPC's and FESCo's
job to fix that, if necessary.
> and I will probably also find that those
> packages have 'accepted' differentiation because no one wants to get
> into a political fight over XYZ package ever again.
And how would you like to fix that? By making packaging guidelines
more lax? By removing the review process, so _every_ packager may
decide whether not to adhere to the guidelines?
Or is it time after several years to revisit the Review Guidelines and
reduce them to a fresh list of most important items packages MUST/SHOULD
> 3) Reviews get lost in limbo for some percentage of people with no way
> of getting them back on track. Then when someone searches for reviews
> to get an idea of how to do them.. those are the open ones that show
Again, what do you want to fix here? Encourage new packagers to follow
the How To Get Sponsored guidelines? Encourage existing packagers to
"swap reviews" with eachother?
A primary problem here also is that people offer extremely niche market
packages, i.e. with a very limited target group. That could mean nobody
else is interested in such packages. Not even users to take notice of
the review request and contribute testing results. Spending time on such
packages would be a hard decision, too. For example, and not limited to
that, I've seen an unfinished remake of a Bard's Tale Construction Set
in the review queue. It may be fun for the submitter to get it included
in the dist, but many a reviewer simply does not consider it worthwhile
to work towards that goal.
> 4) Teaching of how to do basic reviews is "RTFM" with a "Good Luck to
> Find the Right FM" as an unspoken corollary on our mountain of dead
> wiki pages.
> I have seen more than enough IRC conversations of
> A: "I would like to do something, what can I do?"
> B:"Review a package."
> A:"OK how do I review a package?"
> B: "Read through these docs ... url1, url2, url3"
> C: "Oh don't use those URLs they are out of date."
With all due respect, such an example is plain silly and hard
> B: "Oh when did that happen...."
> C: "Oh we decided to drop that part and went to this new system...
> hmmm it says that is obsolete. Oh well someone will know."
> <two days later>
> D: Why didn't anyone tell A to talk to me.
> B,C: Who are you and why would we...
Is it really that bad on IRC these days?
Why don't people mail devel@ anymore these days?
> Or the person goes over to the SIG mailing list or irc group asks a
> question and doesn't get a reply at any time.. then it is usually..
> "Why do we even have this SIG if no one is working on things. Someone
> should do something about that!"
I've never understood the rules about when to create a new mailing-list
and when not. A good rule of thumb is to create a new list only if there's
enough traffic for it. Empty/silent/abandoned lists are scary. Same for
lists with poor/missing descriptions and nobody who fixes it
> So enough whinging. How can we fix this. How do we make it clear which
> pages are to be used to learn how to review a package because Google
> seems to show me different pages depending where I physically search
> from versus the same list every time.
Examples, please. Which pages do you refer to?
> How do we deal with our existing
> mountain of packages to see if they are still in line with reviews?
What does that have to do with the subject ("Sponsor shortage")?
What else would you like to fix? Getting package maintainers to show
that they've taken notice of bug reports in bugzilla even if they don't
respond in any way? Occasionally I report packaging problems but don't
get any response at all.
More information about the devel