Re: PhD and Composer - 2020 approach in the no-mirror era

On Sun, 19 Apr 2020 at 03:39, Maciej Sobaczewski <[email protected]> wrote:

> Hello guys,
>
> long time no see, I know. A lot of things changed around us, I e.g. grew
> up to get my first actual job instead of studying and having way too
> much of a free time :D However, despite how it might look from outside
> I still haven't forgotten about php.net nor my ideas to somehow improve
> its state.
>
> To the point - some of you may remember my move towards migrating PhD
> (PHP's manual builder) to use Composer. Actually, the implementation
> was ready even back then. It worked, I tested it locally multiple times.
>
> If my memory serves me right the main pain point back then was the
> amount of infrastructural changes it would require to make this move.
>
> I'm wondering if now, in 2020, COVID era when some people have a bit
> more time (or rather are forced to sit home and find something to do)
> we would be able to do something more about that. To make it happen.
>
> I'm not even sure about the infrastructure behind building the docs
> right now, AFAIR mirrors or CDN's don't have much to do with it since
> the docs have always been built on single machine and then rsynced to
> the mirrors, right? Has it changed?
>
> The implementation is right here
> https://github.com/php/phd/compare/master...Sobak:composer
> sitting and waiting to be finished. I'm absolutely willing to resolve
> any conflicts with current codebase, do some rebases to clean up the
> history etc. You can count on me when it comes to the code.
>
> The thing I would still need help with would be a cooperation in
> putting that on our servers network. If I'm correct that would mean
> the machine that builts the manual, the editor.php.net machine,
> Windows box that builds CHMs and perhaps euk2.php.net responsible
> for docs.php.net
>
> I lost my access to the last one, which is rather understandable
> given all that time I was not responding etc. Either way, even
> with that access, I still wouldn't be able to finish the migration
> myself.
>
> Straight to the point. Are there people willing and capable of
> finishing the move? Is there even an interest in moving PhD to
> Composer? We still have "php" vendor name reserved by Packagist
> owners. The vendor which is still completely empty.
>
> I'm able to finish the code, bring it up to date and polish it
> within max a week, I think. Yet, without a cooperation it would
> still be a simple branch in the fork. Are there people willing
> to help me?
>
> To be completely fair, I would be more than grateful to finish
> this migration but I'm not going to put my time into polishing
> the code if it's going to stay unused. I absolutely don't want
> to put pressure on anyone as I have no intention and right to
> do so. I just want to be honest.
>
> I'm looking forward you answers and well, it's great to see you
> again, everyone!
>
> Maciej Sobaczewski
>
> PS: I allowed myself to CC the most influencial and active people
> out there (at least the ones that were active during "my times".)
> I hope that you do not take the offence.
>

Hello Maciej,

I don't know much about the infra, but with the migration from SVN
to Git I'd imagine that the online editor would stop working and would
be discontinued.
I also needed to look up what CHM is, and are these still being built?
IIRC the PDF versions of the manual stopped being built a while ago.

I know this is a tangent but a couple of month ago I tried to play
around using pandoc [1] to generate the docs instead oh PhD and it
failed pretty bad as the manual generated by the build uses XML
system entities which pandoc does not support.
What I'm trying to bring forward is that maybe we should look into
refactoring the whole build process so that we can have a build
which doesn't take so much time that we only build the docs and
published them once a day.

Altoutgh this is not what you're are asking, I think it's something
to consider instead of, maybe, spending time on moving PhD to
composer.

Anyway, the effort is highly appreciated from me :)

Best regards

George P. Banyard

[1] https://pandoc.org/