bug#25045: [Website] Packages page takes too long to load

[Alex Sassmannshausen]

Ludovic Courtès writes:

> Luis Felipe López Acevedo <address@hidden> skribis:
>
>> On 2016-11-28 12:00, Alex Sassmannshausen wrote:
>>> Hi Luis,
>>>
>>> Indeed, I had a first bash at solving this problem by providing a
>>> set of
>>> static html pages paginated by the first letter of the package name.
>>>
>>> I'm not particularly wedded to this solution, so if you feel strongly
>>> about going another way, I'd be keen to hear/see about it.
>>>
>>> In the meantime, I'm open to testing/feedback.  I will unfortunately
>>> not
>>> be able to put work into this until at least Saturday/Sunday, as some
>>> Perl work has higher priority at the moment.
>>>
>>> But let me know if you have questions or feedback!
>>>
>>> Best wishes,
>>
>> Hi, Alex :)
>>
>> The solution I had in mind includes an alphabetic index, and
>> pagination as well. However, it includes a few more things, and would
>> take some time to design and implement. So I think we should use your
>> patch to fix the size issue as soon as possible.
>
> I agree!
>
> Alex, was there anything left to address?  If not, feel free to push.
> :-)

Having pushed some other stuff around my calendar, I should now be able
to spend some time on polishing the patch tonight.

>> For what is worth, this is what I had in mind:
>>
>> /packages/
>> First page of the list of packages. Packages listed here only provide
>> a summary: package logo (if any), name, version, description, and an
>> indicator if it has issues. This page also has filters to find
>> packages (for now, alphabetic filter. In the future, category filter,
>> and search box).
>>
>> /packages/page/N/
>> Page N of the list of packages.
>>
>> /packages/a/
>> First page of the list of packages whose name starts with A. Packages
>> are presented the same way as in /packages/.
>>
>> /packages/a/page/N/
>> Page N of the list of packages starting with letter A.
>>
>> /packages/icecat/X.Y.Z/
>> Page with details about IceCat version X.Y.Z. It includes all the
>> information about this package, including its issues (which are
>> currently listed in a separate page along with the issues of other
>> packages: /packages/issues.html). Including the issues of a package in
>> its detail page could avoid having the current issues page grow too
>> much, like the current Packages page.
>>
>> This static pagination and filtering would generate A LOT of pages,
>> but of reasonable size for web browsers to load.

The routing schema looks pretty neat to me! :-)

> Sounds like a good plan as well, though that’s indeed a lot of web pages
> for that rusty CVS repo to handle…
>
> Medium-term, I think we should consider a solution involving pages
> generated on the fly server-side, with a caching proxy (nginx!) in front
> of it.  We’ll have to seek assistance from the gnu.org web masters, but
> ISTR they were not against that idea.

This sounds like a pretty good idea — though I probably won‘t have time
to take the initiative here, I'm happy to do review and discussion of
patches.

Alex

>
> Thanks!
>
> Ludo’.