On Sat, Aug 4, 2018 at 3:25 PM, Miro Hrončok
> On 4.8.2018 12:23, Greg Sheremeta wrote:
>> On Sat, Aug 4, 2018 at 3:48 AM Miro Hrončok
>> <mailto:firstname.lastname@example.org>> wrote:
>> On 4.8.2018 01:17, Greg Sheremeta wrote:
>> > Hi,
>> > This page
>> > is terribly outdated. Even when it was created years ago, IMO the
>> > was questionable. Today, it's definitely bad advice.
>> > Modern web
>> > injected. I realize bundling libraries is bad for an old-school
>> (try to
>> > find react and friends), and the page is leading to confusion on
>> my team.
>> > To prevent
confusion, acceptable options would be: either simply
>> > deleting the page, or placing a giant "don't follow this
>> > advice" banner at the top.
>> We don't generally do either of those. If the
>> they need to to be updated, not deleted.
>> Ok. Then I suggest this
page be updated to roughly say client-side
> Feel free to propose a ticket
> Best tickets include:
> * draft guideline on the wiki and diff
> * rationale (explained in detail, so even nonexperts in the given domain
> (here JS) could get the reasoning behind it)
> As for now, I personally don't like your idea, as it
> about what should the packager do if they need to package a web
> that uses client side JS.
> For example: the app ships
client side JS bundled, but minified.
> Or: the app expects me to run npm install to get the client side JS.
It looks like he wants webpacks, instead of RPMs. This has been tried
before, with CPAN, pip, ant, maven, and gradle, and they run into the
same issues. They're very language specific, they're bad at avoiding
conflict with other installed software, they're bad at auditing or
reporting what contents deposited on your filesystem come from what
application. Those issues are not all the fault of the packaging
system: much of it is the fault of software authors who cannot follow
conventions about how to name and where to install their software.
Frankly, the enforced rules and hard-won lessons of RPM bundling have
helped tremendously. Some tools, like cpan2rpm and py2pack, have even
gotten useful for RPM bundling packages from such software building
and bulding tools.
It sounds like, if webpacks are going to be supported for RPM
packages, then it needs wrappers to pack them into RPM bundles.
Perhaps something like "cpan2rpm" could be done for webpacks?
It is a similar situation to previous languages. However, it's worse here
isn't going to scale.