Freywa wrote:The attachment to this post contains all syntheses of oscillators with LifeWiki pages in this thread before the start of 2014 that are not on LifeWiki and which have not been superseded by later substantial improvements. However, because incremental improvements can take time, I've only zipped up the most efficient complete synthesis of the objects chosen. A view to finding any further improvements and then creating the relevant synthesis files on LifeWiki is taken.
Thanks for doing this. I've uploaded the two p37s manually to the LifeWiki, but the process is fairly time-consuming so I won't continue immediately. Maybe the rest could be done via the auto-upload script.
The 44-glider synthesis for
Beluchenko's other p37 was actually improved to 40 gliders a few posts down, so I've manually added that synthesis and an extra link to the _synth file.
It looks like the comments are pretty good on all the attached files already, so they can be added as they are to the RLE namespace, and the auto-upload script will take care of (almost) everything from there. I've added the synth pattern for 56p27 as an example. Basically the idea is just to type "RLE:56p27_synth" (pname all lowercase, with "_synth" appended) into the LifeWiki search bar, create the file, and paste in the text from the attachment:
http://conwaylife.com/w/index.php?title ... ction=edit
Then it's just a matter of going to the
53P27 article and adding the following after "rle = true" in the infobox parameters:
Code: Select all
|synthesis = 52
|synthesisRLE = true
More Stuff That Needs Doing
If anyone would like to volunteer to do the above RLE namespace edits for the remaining 21 files in the attachment, that would be great. The auto-upload script would then
- automatically create an RLE file with the correct name, 46p27_synth.rle, ready for upload;
automatically add a #N header line to the RLE -- #N 56p27_synth.rle, in this case;
- automatically add a comment with a link to the LifeWiki article for the pattern being synthesized.
- automatically add a comment with a link to where this same RLE file will be stored on the LifeWiki server.
When there are a lot of these auto-created pattern or synth files, they can be sent to Nathaniel to do a bulk upload. However, now that the auto-upload script is close to being caught up and there aren't hundreds of files to handle, it's fairly easy for any LifeWiki moderator to run quickly through a folder of a few dozen new uploadable files and drop them onto the server one by one (the only available method for non-Nathaniels).
I wrote the auto-upload script just because it ends up being enormously more time-consuming if every file has to be hand-edited, renamed, gliders counted, etc. before the upload happens. All that tedious work ends up being a task to be avoided at all costs -- which is presumably why nobody has done it in a lot of cases, and synthesis files have fallen this far behind current reality.
Room for Improvement
One downside is that adding "synthesisRLE = true" is a temporary lie: the file won't actually be there on the server until the next time the auto-upload script is run. To me it seems easier to do the editing just once when the synth pattern is added to the RLE namespace, and allow reality to lag behind the promise a little bit.
-- I do admit that that's a bit suboptimal, though. Would it be preferable if the auto-upload script created a report of all the cases where rle=true or synthesisRLE=true tags are missing from the articles, so that someone could go through and do those updates systematically later?
Currently the auto-upload system doesn't pay attention to "rle = true" or "synthesis = true" or "plaintext = true" tags -- it just checks for pattern, _synth, and .cells files on the server, and creates any missing files that it can figure out how to create using whatever is in the RLE namespace. It could easily make a list of articles that will need any of the above parameters added to them, once the auto-upload process is complete.
The system is currently conservative about uploads; it doesn't prepare a pattern or synth file for upload if there's a file already uploaded. So no files on the server will ever be automatically overwritten. This wouldn't change with the proposed adjustment to the script.
Any other suggestions, comments or complaints on the current (evolving) status quo for synthesis uploads?
TL;DR: Adding latest-and-greatest synthesis patterns to the RLE namespace -- especially with reference names, dates, URLs and glider counts in the comments -- is now a good step forward in getting the LifeWiki up to date on known syntheses. Maybe it can become a bigger step forward with the next iteration of the auto-upload script.
As discussed
elsewhere, it doesn't seem to make sense at this point to use Mark Niemiec's 2015 synthesis database as a source for patterns to be uploaded to the LifeWiki. It's just a bit too likely that the newest cheapest synthesis won't be there, but it will hopefully be in the
upcoming 2019 release.