talk me beautifull
category: general [glöplog]
There's almost a thousand YouTube links that are broken (a random example is this prod). I have copies of all of them in my archive, but from experience, just linking to those as “youtube” isn't accepted. What do we do?
We could remove them, but that seems silly when we actually have a copy.
We could re-upload them to YouTube, but I honestly don't want a bazillion demoscene captures on my channel, so it would have to be someone else's (and they'd need to care about stuff like metadata).
We could downgrade them to “video” and link to the file on the archive (this) or the super-simple player (this, though that's not a “video”). Or we could put it on a different host, or on some lost-and-found directory on scene.org.
I don't have any strong opinions, except that I'm not going through all of these videos manually. It has to be something automated or semi-automated.
We could remove them, but that seems silly when we actually have a copy.
We could re-upload them to YouTube, but I honestly don't want a bazillion demoscene captures on my channel, so it would have to be someone else's (and they'd need to care about stuff like metadata).
We could downgrade them to “video” and link to the file on the archive (this) or the super-simple player (this, though that's not a “video”). Or we could put it on a different host, or on some lost-and-found directory on scene.org.
I don't have any strong opinions, except that I'm not going through all of these videos manually. It has to be something automated or semi-automated.
not yet .
Quote:
not yet .
Please stop it. You are not being useful, and you are not being funny.
useful ? miss one L no ?
, so you judge , you think you have rights to judge someone from your own race ?
funny no but a lesson from Aubade could or should be a good idea .
holidays
12c4.
, so you judge , you think you have rights to judge someone from your own race ?
funny no but a lesson from Aubade could or should be a good idea .
holidays
12c4.
I still don’t understand why we can’t just have a “repository” link type that doesn’t mandate a particular code hosting platform…
"fragmentation" doesn't matter in the least, and it is very silly to ban alternatives to github just because github has a monopoly. because to the person clicking the link, it does not matter if it's github or a better website, since the website provides the same functionality to them anyway (except with a higher likelyhood of being up at the time)
as stated before, it's up to y'all to make a convincing argument what other repository platform deserves to be supported. misrepresenting arguments that others make probably isn't the best way to do that.
perhaps it would help clear up some confusion if we understood what supporting a new repository platform entails from the pouet staff side. if pouet were to support codeberg, for example, what's the actual work involved beyond just adding it to the list in the FAQ?
Quote:
it's up to y'all to make a convincing argument what other repository platform deserves to be supported.
From what I can see, a bunch of people have already answered that with “all of them, because the source code link is valuable information regardless of whether it’s on a well known site or not” and are now waiting for an explanation of why that’s not a convincing argument.
I *think* the answer to that is: because if we support a limited number of well-known sites, we can do pattern-matching on the URL to get to the actual downloadable repo, and then someone like Sesse can mirror it as a guard against the site going away. But please correct me if I’m misrepresenting that…
If so, I might suggest adding gitlab and codeberg as accepted hosts.
Quote:
I *think* the answer to that is: because if we support a limited number of well-known sites, we can do pattern-matching on the URL to get to the actual downloadable repo, and then someone like Sesse can mirror it as a guard against the site going away. But please correct me if I’m misrepresenting that…
I'll be happy to mirror whatever becomes common; I just added matching against Bitbucket URLs today because it seems to be in use for some prods already. I don't have a problem with people adding more sites for my part. The worst thing that can happen from my perspective is that people add links to some site that I don't support, and the real alternative is surely that the link isn't added at all, which is no better from a preservation point of view?
In any case, the current policy predates me (or anyone else, to the best of my knowledge) mirroring GitHub repositories from Pouët links at all.
(I don't have any role in scene.org nor Pouët, so don't take this as an official position for anyone except myself)
gasman: well that would imply that we intend to restore those backed up repositories in case a site goes away. i'm not necessarily sure that's a good idea.
furthermore, explanations why some specific sites are not (yet) supported are on the previous pages of this thread, i'm not going to reiterate that every time ppl ask essentially the same question again
furthermore, explanations why some specific sites are not (yet) supported are on the previous pages of this thread, i'm not going to reiterate that every time ppl ask essentially the same question again
I've tried reading through the arguments against in the thread, and it strikes me that it consistently talks about a codeberg label. But I don't think anyone's arguing for giving Codeberg its own label, just allowing Codeberg URLs? So I'm a bit confused; is everyone just talking past each other?
Quote:
I've tried reading through the arguments against in the thread, and it strikes me that it consistently talks about a codeberg label. But I don't think anyone's arguing for giving Codeberg its own label, just allowing Codeberg URLs? So I'm a bit confused; is everyone just talking past each other?
Indeed, I never argued for a codeberg label, but am very annoyed (along with e.g. yx) that codeberg URLs are now also banned.
Quote:
I *think* the answer to that is: because if we support a limited number of well-known sites, we can do pattern-matching on the URL to get to the actual downloadable repo, and then someone like Sesse can mirror it as a guard against the site going away. But please correct me if I’m misrepresenting that…
If so, I might suggest adding gitlab and codeberg as accepted hosts.
I have also answered this before: because git uses a standard protocol for downloading whole code repositories (instead of saving a snapshot of the HTML of the code hosting site, which is kind of useless), this should work for any code hosting site: add ".git" to the end of the URL, and fetch it as a git repository (i.e. literally just do a "git clone https://hostingsite.tld/username/reponame.git"). This is exactly the reason why maintaining a list of one or two allowed websites is just plain silly.
Merci beaucoup Havok
.
J'aimerai faire un logo pouet mais je ne sais comment faire pour en faire une animation...
et je ne sais comment modifier la couleur du fond de pouet , en modifiant bien sur avec fader les anciens logos en fesant une forme de croix certes , mais qui au dond ne sont que des + .
Bref , j'ai besoin a minima , par mail , sinon le logo sera sur mon facebook . pour une version noire ou grise de pouet .
Bonne soirée Havok.
o.bechard@gmail.com
.
J'aimerai faire un logo pouet mais je ne sais comment faire pour en faire une animation...
et je ne sais comment modifier la couleur du fond de pouet , en modifiant bien sur avec fader les anciens logos en fesant une forme de croix certes , mais qui au dond ne sont que des + .
Bref , j'ai besoin a minima , par mail , sinon le logo sera sur mon facebook . pour une version noire ou grise de pouet .
Bonne soirée Havok.
o.bechard@gmail.com
Is there any way to “fast-track” rescues of dead links that don't have existing mirrors?
As an example, this prod is hosted on asd.gr, which is broken according to cardboard (certificate issues). My mirror has a copy; if scene.org had a copy, I'd just send an edit request, but it doesn't (nor any other working URL that I know of).
scene.org has an ASD directory, but to put it there, I would need to first download from my site, go into ftp.scene.org/incoming, mkdir some directories, upload the .zip file, write a .txt file to go with the upload, and then wait for a scene.org operator to have time to moderate it. I don't get a notice, so I have to remember that I did this and keep polling ftp.scene.org to check if the file has been accepted and moved, and only then send a Pouët change request. If this were for a group that doesn't already have a directory, I guess there's also more process? Or if the group even isn't known (e.g., this prod is broken), it usually goes into some gloperator's scene.org directory.
Is there a way to make this smoother?
As an example, this prod is hosted on asd.gr, which is broken according to cardboard (certificate issues). My mirror has a copy; if scene.org had a copy, I'd just send an edit request, but it doesn't (nor any other working URL that I know of).
scene.org has an ASD directory, but to put it there, I would need to first download from my site, go into ftp.scene.org/incoming, mkdir some directories, upload the .zip file, write a .txt file to go with the upload, and then wait for a scene.org operator to have time to moderate it. I don't get a notice, so I have to remember that I did this and keep polling ftp.scene.org to check if the file has been accepted and moved, and only then send a Pouët change request. If this were for a group that doesn't already have a directory, I guess there's also more process? Or if the group even isn't known (e.g., this prod is broken), it usually goes into some gloperator's scene.org directory.
Is there a way to make this smoother?
@sesse: i don't know how or if something like that can become smoother. that technical stuff is not my field. but in the end someone has to upload it. and if a group does not have a directory it gets created.
(and i just assume you talk about the lost_found_and_more directory. this was the archive i build over the years which was on untergrund, then on my webspace and then after problems with https it got moved (after getting convinced by menace) to scene.org. and then gargaj replaced the domain-part in the links at pouet.
(and i just assume you talk about the lost_found_and_more directory. this was the archive i build over the years which was on untergrund, then on my webspace and then after problems with https it got moved (after getting convinced by menace) to scene.org. and then gargaj replaced the domain-part in the links at pouet.
Well, I thought maybe there could be a way to skip moderation if the file had a known SHA256 or similar (I assume the main point of moderation is to keep illegal stuff out of the archive). I don't know the inner workings of scene.org, though, neither technically nor organization-wise.
Of course, I could make my own lost-and-found.sesse.net and put stuff there, but then we could just as well just link to pouet-mirror.sesse.net and I believe people don't want that due to redundancy concerns.
Of course, I could make my own lost-and-found.sesse.net and put stuff there, but then we could just as well just link to pouet-mirror.sesse.net and I believe people don't want that due to redundancy concerns.
that ASD backup is a broken zip file, and ITS is definitely on scene.org somewhere, or at least it used to be, the main problem seems to be that it's probably a different version (or else you would have found it by comparing CRC's i suppose)
so skipping moderation at one point would probably introduce a need for moderation somewhere else, or to put it differently, we need known *good* SHA's and that'll take plenty f manhours
so skipping moderation at one point would probably introduce a need for moderation somewhere else, or to put it differently, we need known *good* SHA's and that'll take plenty f manhours
Yeah, I should have remembered that the ASD prod was an outlier, it was a bad example. And yes, if the contents are the same, I will find the file; sometimes there's slightly different versions with updates and finals and .nfos and stuff (and since I almost never recrawl URLs that actually succeeded, I don't get updates when people silently change their files).
FWIW, even now with manual moderation, there's definitely dupes being uploaded to scene.org, so I don't think that would change. (I can give you a pretty long list of files in lost_found_and_more that are matches of other stuff, but it's not that easy just to remove in case someone is linking to them.)
FWIW, even now with manual moderation, there's definitely dupes being uploaded to scene.org, so I don't think that would change. (I can give you a pretty long list of files in lost_found_and_more that are matches of other stuff, but it's not that easy just to remove in case someone is linking to them.)
Quote:
ftp:// links as main download have been all but rooted out completely
I noticed today there are many more of them than we actually knew about, because certain sites love redirecting from http to ftp. E.g., there would seem to be hundreds of links on noname.c64.org which are just a simple redirect to ftp://ftp.scs-trc.net or other places, which basically means the download link is broken in most browsers. (I don't have an exact count, because I only recently started saving redirect information during crawl.)
Is there a policy of peeling away thin redirects to leave the real URL in the download? Or do these stay invisible?
There's no such policy but it does sound like an undesirable situation so I'll add this to my longer term to-do list. Do you perhaps already have some list of affected prods?
I have a list of ~1000 such prods (most of them CSDB links), but that would only be the tip of the iceberg since I didn't start collecting this information until a couple of days ago. status.json carries redirect information now, so you can grab it from there.
I have no idea what cardboard's policies are in general (I don't think the source is available?), but you would imagine it could also collect this kind of information when it re-crawls. Although cardboard is honestly a bit of a black hole to me; I think maybe there are some sites it doesn't validate at all or very rarely, but I don't think I've figured it out fully.
I have no idea what cardboard's policies are in general (I don't think the source is available?), but you would imagine it could also collect this kind of information when it re-crawls. Although cardboard is honestly a bit of a black hole to me; I think maybe there are some sites it doesn't validate at all or very rarely, but I don't think I've figured it out fully.
(There's also probably a fair amount of simple http -> https redirects that just should be incorporated on Pouët's side without too much deliberation.)