Also tried setting you as author... can you complete an Edit ok ? (I haven't purged permissions yet... curious if needed...)
The OB# sort thing; yeah I can see that's something known about that's just a limitation of the plugin. Can be solved quite simply (though not quickly) with approval in the future - probably need to badger someone at next p2lf with a list of things
For now... I've removed the sort on the ID column to remove the confusion. I had a quick look at adding the object date column too (as opposed to the post-imported date column), but that's also in the simple but more than 5 mins throwing-spanners-without-any-thought category
BTW... in the post content you should be able to use the triple backticks to include code blocks,- maybe handy for objects that are just snippets or tips.
The code lang should be auto-detected. Won't show up well in the markdown preview box, but will work out when posted.
I thought we had the keyboard symbols encoding too, but seems that broke somewhere. Poop. It'll come one sunny day! That might be a quick thing to tweak.
@VonSzarvas said:
Also tried setting you as author... can you complete an Edit ok ? (I haven't purged permissions yet... curious if needed...)
Yep, editing objects still reverts author field.
Also, I looked at the corrupted tunes.zip... I think everything that was an LFS object in the git repo got borked on import. Might have to do it all again.
As for the child-zips, yeah, that's annoying.
Looks like downloading a GitHub repo zip archive to windows gets the files intact. Whereas downloading the GitHub repo zip archive to linux via wget appears to only bring in the git-lfs headers.
I could understand the difference when using git-clone without having git-lfs installed (though git-lfs was installed on the downloading server), but for wget to bork seems almost impossible; worthy of closer analysis.
@Wuerfel_21
Editing surely now solved. Could you edit and re-add your files please. (Security settings won't let me do that for you, and those will need some time to adjust for me to do that).
I wonder, if filter by Author should be made possible?! It is interesting to look for things gifted people like PhiPi have given us. Also I do think, that Parallax should honour the Authors. They have invested a lot of time, so show their name instead of "Archiver"!
I must say, that I was very much shocked, that Parallax did not realize the big treasure of the Obex for so long and also the treasure of the first forum, where so much was lost also. Good to see some different sign now!
I wonder, if filter by Author should be made possible?! It is interesting to look for things gifted people like PhiPi have given us. Also I do think, that Parallax should honour the Authors. They have invested a lot of time, so show their name instead of "Archiver"!
I must say, that I was very much shocked, that Parallax did not realize the big treasure of the Obex for so long and also the treasure of the first forum, where so much was lost also. Good to see some different sign now!
Today, I had the chance of downloading a couple of objects from the new obex.
The first weird thing I noticed is that archives are downloaded with the code number name: OB2841.zip, OB2877.zip. Is this right ? I don't think it is right... how do I know what I have downloaded ?
The more annoying thing is that of 3 objects I have downloaded today, two have errors (missing object files or errors in the source itself). I don't think this is good.
I was wondering if it is possible to implement a validation process. Now that we have a number of compilers available that can be started from the command line, would be good if there is a way to validate uploaded objects to make sure that they can actually compile, notify the author if something is missing, notify when compiler changes breaks existing code (this is related to the recent new keywords discussion in the PNut thread). More or less a sort of "continuous integration" process that makes sure things doesn't break.
@macca said:
The first weird thing I noticed is that archives are downloaded with the code number name: OB2841.zip, OB2877.zip. Is this right ? I don't think it is right... how do I know what I have downloaded ?
What would you prefer ?
Maybe a sanitized version of the title, perhaps with spaces swapped to underscores and with a reasonable max-length string limit ?
If so, would you prefer the OB number kept as a prefix, or just go with the title?
(or something completely different?)
@macca said:
The first weird thing I noticed is that archives are downloaded with the code number name: OB2841.zip, OB2877.zip. Is this right ? I don't think it is right... how do I know what I have downloaded ?
What would you prefer ?
Maybe a sanitized version of the title, perhaps with spaces swapped to underscores and with a reasonable max-length string limit ?
If I remember correctly, with the old obex the archives had the same name as when uploaded, but yes, a sanitized version of the name would be good to remember what was downloaded.
I don't think that spaces and name length would cause problems with modern OSes.
@macca said:
I don't think that spaces and name length would cause problems with modern OSes.
Oh they do. Spaces are the number one enemy of shell scripts everywhere. And windows still has the same old 255 character path limitation. The actual OS supports more, but it's limited by default to avoid creating paths that break old software.
@macca said:
If I remember correctly, with the old obex the archives had the same name as when uploaded, but yes, a sanitized version of the name would be good to remember what was downloaded.
I don't think that spaces and name length would cause problems with modern OSes.
You might be right, but I don't have access to those original files.
Spaces and length - I'd rather set something else we are tempting fate, and we need to sanitize out a bunch of chars anyway.
Will the Git repo continue to be updated? I found it very convenient - just run git pull and have the latest versions of everything at my fingertips. Would it be possible to make the new PLX HUB get periodically synchronized with the Git repo, or, even better, make it a frontend for the Git repo that hides all the Git details?
This obnoxious cookie thing has appeared and it's broken on dark mode:
(Do you even need this if you don't have any tracking cookies? I think normal login cookies don't need consent like this because you already agreed when creating an account to log in with)
Heading uses bad font definition
Never just use a font that the user may or may not have installed. Always at least add one of the generic fallbacks like sans-serif.
The normal body text also uses fonts that I don't actually have installed, but that at least has a sans-serif fallback so it doesn't look silly.
Wow, yeah, that default font setup was curious in places on that template. Also, the cookie monster removed, and some author names added to the table (where the github readme had that info in a recognizable style - not all did and will require manual fixing).
ps. Before anyone asks, the author thing was a quick visual patch. Making the imported author names searchable/sortable is on the todo list, along with other search/sort options like the OB#, last date, and a whole bunch of testing and improving, etc... I'd like to add a button at some point, for registered users to "claim their imported objects". Once I get permission to proceed it will happen quickly.
Many features are in default state, not final state - still trying all the options!
The search box at the top should find just about anything, and most columns are sortable.
The filters are at the bottom of each column at the moment (though they will probably be moved up top later!)
Direct download buttons and category links haven't been added yet.
Total number of objects restricted on this table whilst testing.
Dark theme not configured, at least two double-darks lurkin'
Feels like this table is more capable and a better direction. What say ye all ?
BTW... The previous table remains operational, and all objects submitted by users are safe from changes!
Feel free to contribute objects any time, or let us know of reasons you can't if there are features you need that haven't appeared yet.
The page selector is rubbish without any way to specify a page number. The uselessness of that style was never a problem historically because it was always easy enough to just specify the page number in the URL ... But by using scripting it doesn't work that way now.
If anyone wishes to post a code snippet to OBEX without attached files, then the same code format as here on the forum can be used, and language should be automatically recognised.
Ie. Surround the code with three backticks, on the line before and also the line after the code.
@evanh said:
The page selector is rubbish without any way to specify a page number. The uselessness of that style was never a problem historically because it was always easy enough to just specify the page number in the URL ... But by using scripting it doesn't work that way now.
I'll take a look and see what options exist to improve on that.
@evanh said:
The page selector is rubbish without any way to specify a page number. The uselessness of that style was never a problem historically because it was always easy enough to just specify the page number in the URL ... But by using scripting it doesn't work that way now.
I'll take a look and see what options exist to improve on that.
Seems like that's not a standard feature, so best left until we have the new table fully fleshed out. (Although I've added some page-navigation options that might help)
In the meantime, this search URL exists, and accepts most search terms including author name or content:
Comments
OK, here's the gremlin in the github index:
I don't have permission to fix that github, but maybe I can fake something locally....
Lunchbreak update: Crud fixed, opnA imported. HOORAH !
Also tried setting you as author... can you complete an Edit ok ? (I haven't purged permissions yet... curious if needed...)
The OB# sort thing; yeah I can see that's something known about that's just a limitation of the plugin. Can be solved quite simply (though not quickly) with approval in the future - probably need to badger someone at next p2lf with a list of things
For now... I've removed the sort on the ID column to remove the confusion. I had a quick look at adding the object date column too (as opposed to the post-imported date column), but that's also in the simple but more than 5 mins throwing-spanners-without-any-thought category
BTW... in the post content you should be able to use the triple backticks to include code blocks,- maybe handy for objects that are just snippets or tips.
The code lang should be auto-detected. Won't show up well in the markdown preview box, but will work out when posted.
I thought we had the keyboard symbols encoding too, but seems that broke somewhere. Poop. It'll come one sunny day! That might be a quick thing to tweak.
Yep, editing objects still reverts author field.
Also, I looked at the corrupted tunes.zip... I think everything that was an LFS object in the git repo got borked on import. Might have to do it all again.
editing should be solved now.
As for the child-zips, yeah, that's annoying.
Looks like downloading a GitHub repo zip archive to windows gets the files intact. Whereas downloading the GitHub repo zip archive to linux via wget appears to only bring in the git-lfs headers.
I could understand the difference when using git-clone without having git-lfs installed (though git-lfs was installed on the downloading server), but for wget to bork seems almost impossible; worthy of closer analysis.
@Wuerfel_21
Editing surely now solved. Could you edit and re-add your files please. (Security settings won't let me do that for you, and those will need some time to adjust for me to do that).
-- yes, to change author, I know
Seems to work now
Phew. Thxs.
Hi,
good, that there is Obex again!
I wonder, if filter by Author should be made possible?! It is interesting to look for things gifted people like PhiPi have given us. Also I do think, that Parallax should honour the Authors. They have invested a lot of time, so show their name instead of "Archiver"!
I must say, that I was very much shocked, that Parallax did not realize the big treasure of the Obex for so long and also the treasure of the first forum, where so much was lost also. Good to see some different sign now!
Christof
+1
Today, I had the chance of downloading a couple of objects from the new obex.
The first weird thing I noticed is that archives are downloaded with the code number name: OB2841.zip, OB2877.zip. Is this right ? I don't think it is right... how do I know what I have downloaded ?
The more annoying thing is that of 3 objects I have downloaded today, two have errors (missing object files or errors in the source itself). I don't think this is good.
I was wondering if it is possible to implement a validation process. Now that we have a number of compilers available that can be started from the command line, would be good if there is a way to validate uploaded objects to make sure that they can actually compile, notify the author if something is missing, notify when compiler changes breaks existing code (this is related to the recent new keywords discussion in the PNut thread). More or less a sort of "continuous integration" process that makes sure things doesn't break.
What would you prefer ?
Maybe a sanitized version of the title, perhaps with spaces swapped to underscores and with a reasonable max-length string limit ?
If so, would you prefer the OB number kept as a prefix, or just go with the title?
(or something completely different?)
If I remember correctly, with the old obex the archives had the same name as when uploaded, but yes, a sanitized version of the name would be good to remember what was downloaded.
I don't think that spaces and name length would cause problems with modern OSes.
Oh they do. Spaces are the number one enemy of shell scripts everywhere. And windows still has the same old 255 character path limitation. The actual OS supports more, but it's limited by default to avoid creating paths that break old software.
You might be right, but I don't have access to those original files.
Spaces and length - I'd rather set something else we are tempting fate, and we need to sanitize out a bunch of chars anyway.
Will take a look at it.
Will the Git repo continue to be updated? I found it very convenient - just run
git pulland have the latest versions of everything at my fingertips. Would it be possible to make the new PLX HUB get periodically synchronized with the Git repo, or, even better, make it a frontend for the Git repo that hides all the Git details?At the moment I'm not aware of plans to continue git repo. Periodic sync feature not too tough to implement so adding to the list for Plx to consider
Parallax Hub is now Parallax OBEX...
https://obex.parallax.com
Ready for all your lovely objects
Just realised the scripting Smile doesn't create links. No way to post a search or parametric result here.
??
A link to a specific list of objects. Ie: So others can see the same list as me.
Not a biggie, linking to individual objects will do.
So, uh, some breakage seemed to have happened:
This obnoxious cookie thing has appeared and it's broken on dark mode:
(Do you even need this if you don't have any tracking cookies? I think normal login cookies don't need consent like this because you already agreed when creating an account to log in with)
Heading uses bad font definition
Never just use a font that the user may or may not have installed. Always at least add one of the generic fallbacks like
sans-serif.The normal body text also uses fonts that I don't actually have installed, but that at least has a sans-serif fallback so it doesn't look silly.
Wow, yeah, that default font setup was curious in places on that template. Also, the cookie monster removed, and some author names added to the table (where the github readme had that info in a recognizable style - not all did and will require manual fixing).
ps. Before anyone asks, the author thing was a quick visual patch. Making the imported author names searchable/sortable is on the todo list, along with other search/sort options like the OB#, last date, and a whole bunch of testing and improving, etc... I'd like to add a button at some point, for registered users to "claim their imported objects". Once I get permission to proceed it will happen quickly.
Been experimenting with a new table that allows for the extra sorting and filtering of author and object number fields (among other things).
Logged in users can see the progress here : https://obex.parallax.com/obex-testtable/
Many features are in default state, not final state - still trying all the options!
Feels like this table is more capable and a better direction. What say ye all ?
BTW... The previous table remains operational, and all objects submitted by users are safe from changes!
Feel free to contribute objects any time, or let us know of reasons you can't if there are features you need that haven't appeared yet.
The page selector is rubbish without any way to specify a page number. The uselessness of that style was never a problem historically because it was always easy enough to just specify the page number in the URL ... But by using scripting it doesn't work that way now.
PS: The sort buttons on the test page are working, thanks. They do good!
If anyone wishes to post a code snippet to OBEX without attached files, then the same code format as here on the forum can be used, and language should be automatically recognised.
Ie. Surround the code with three backticks, on the line before and also the line after the code.
I'll take a look and see what options exist to improve on that.
Seems like that's not a standard feature, so best left until we have the new table fully fleshed out. (Although I've added some page-navigation options that might help)
In the meantime, this search URL exists, and accepts most search terms including author name or content:
https://obex.parallax.com/posts-search/?ptp_search=motor
https://obex.parallax.com/posts-search/?ptp_search=evanh
Nice. That might just do exactly what I've been wanting for future linked searches. Time will tell as the database is fleshed out.