MarbleousDave
He/Him
Player (13)
Joined: 9/12/2009
Posts: 1559
There seems to be a little problem with the movie counter. 1916 was skipped for no reason.
Post subject: Publication number 1916 skipped
Editor, Experienced player (570)
Joined: 11/8/2010
Posts: 4036
PikachuMan wrote:
There seems to be a little problem with the movie counter. 1916 was skipped for no reason.
(Quote from here.) Why was publication 1916 skipped? The publications just go 1914, 1915, 1917...
Post subject: Re: Publication number 1916 skipped
Emulator Coder, Skilled player (1113)
Joined: 5/1/2010
Posts: 1217
CoolKirby wrote:
Why was publication 1916 skipped? The publications just go 1914, 1915, 1917...
Basically, publishing 1916M hit the publication bug and the attempts to repair messed up the database. While hitting publication bug would not ordinarily cause publication number to be skipped (at least 3 such publications have been successfully fixed), Grunt thought a good way to fixing it would be to unpublish the movie and then republish. Unfortunately, doing this got the site database into some weird state where the entry for 1916M both existed (attempts to add it were rejected as duplicate) and didn't exist (attempts to look it up returned row not found) at the same time. Thus, the movie was republished as 1917M and 1916M was left as a gap.
Editor, Experienced player (570)
Joined: 11/8/2010
Posts: 4036
OK, that makes sense. So theoretically a movie could be unpublished without recoding the site? I thought it was impossible.
Emulator Coder, Skilled player (1113)
Joined: 5/1/2010
Posts: 1217
CoolKirby wrote:
So theoretically a movie could be unpublished without recoding the site? I thought it was impossible.
The site interface doesn't allow removing publications, that requires directly mucking with the database.
Joined: 11/22/2004
Posts: 1468
Location: Rotterdam, The Netherlands
PikachuMan wrote:
There seems to be a little problem with the movie counter. 1916 was skipped for no reason.
That might have been because this is the first Windows submission, and something went wrong while updating the site to include that as a category. So 1916 would be the broken publication that later got deleted. edit: see http://tasvideos.org/forum/viewtopic.php?t=12069 edit: okay, so apparently this post of mine got merged into another topic. It used to be in another topic where the answer hadn't already been given, so that's why I'm stating the obvious here. (I'm not normally that daft, honest.)
Editor, Emulator Coder, Expert player (2157)
Joined: 5/22/2007
Posts: 1134
Location: Glitchvania
I am still wondering about why Movie #666 was skipped.
<klmz> it reminds me of that people used to keep quoting adelikat's IRC statements in the old good days <adelikat> no doubt <adelikat> klmz, they still do
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
If it were up to me I would suddenly jump to movie in whim just to stop this kind of questions (or make them seem insignificant in comparison) ;-) There is a general development principle that database-internal numbers should not be exposed in public interface. According to that principle, the URLs referring to movies should be like /TAS/SMB1byKlmz/v1 rather than /1330M.html . But the fact is that these numbers are a handy way to refer to individual movies. I would just prefer that people take it as that, a handy way to refer to individual movies, without trying to assign any particular meaning to those numbers; their magnitude or their relative ordering. Yes, the database assigns the number with a certain algorithm, in a successive sequence, but the numbers may also skip, and have irregular sequences depending on a (small) number of issues.
Banned User
Joined: 3/10/2004
Posts: 7698
Location: Finland
If the numbering were strict, it would convey useful information: The number would directly tell how many movies have been published up to the specified one. (Hence if you look at the newest publication, you'll know how many movies have been published during the history of the site.) Hence the desire for the number to be significant in such a way is understandable.
Editor, Experienced player (570)
Joined: 11/8/2010
Posts: 4036
klmz wrote:
I am still wondering about why Movie #666 was skipped.
I think it should be obvious that the site was coded to skip it. Would you want one of your movies to be publication number 666? Or the skip was just a random server error. Also, I agree with Warp. I used to think that the publication numbers told how many publications were really on the site, and I would try to figure out why new submissions get much higher numbers than new publications. It would be nice to know how many movies have been published since the site's beginning, but I didn't even know these errors existed, or how often they occurred.
Joined: 11/22/2004
Posts: 1468
Location: Rotterdam, The Netherlands
Bisqwit wrote:
There is a general development principle that database-internal numbers should not be exposed in public interface.
There are some advantages to doing it, though. It simplifies URLs, for one thing. The Internet Archive's URLs can be pretty messy and long. And of course it's much easier to look something up in a database when you have some ID to refer to. In our case it probably wouldn't matter too much, performance wise, because we have only thousands of entries (and not millions of entries, like comments on Reddit for example), but it helps.
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
Dada wrote:
There are some advantages to doing it, though. It simplifies URLs, for one thing. The Internet Archive's URLs can be pretty messy and long. And of course it's much easier to look something up in a database when you have some ID to refer to. In our case it probably wouldn't matter too much, performance wise, because we have only thousands of entries (and not millions of entries, like comments on Reddit for example), but it helps.
No, for the purpose of shortening URLs one should create a shortened URL scheme, spefically linking "this corresponds to this". For example, in one of the projects developed at where I work, both this and this are URLs created with an algorithm that does not expose database-internals. If some day the platform of implementation changes, or the database layout is changed radically, the URLs will still continue to be valid, because they do not expose anything that is in any way tied to how the database works. What we have at TASVideos is just me having been lazy.
Banned User
Joined: 3/10/2004
Posts: 7698
Location: Finland
I think you are equivocating movie publication numbers with "database internals". If I understand correctly, when you use the term "database internal" you are referring to data in the database that's just used internally for bookkeeping, indexing and so on, and being completely internal, this data could well change at any moment without the external behavior of the site changing in any way. In other words, it's just an internal implementation detail, irrelevant to the outside. However, I don't consider movie publication numbers mere "internal details". They convey (assuming they were consistent) interesting and useful information. They tell if a given publication is newer than another publication, they tell how many publications there have been so far, and they provide a simple ID which can be used to refer to a specific publication easily and, more importantly, unambiguously. (You can say things like "check movie number 1234 for an example" and it will be short, completely unambiguous, and handy.) In fact, publication numbers are probably the easiest way of referring to specific publications. I can't think of any easier way.
Joined: 2/26/2011
Posts: 98
Bisqwit wrote:
If it were up to me I would suddenly jump to movie in whim just to stop this kind of questions (or make them seem insignificant in comparison) ;-)
Wow. Your gif worked perfectly on me. You must use the exact same ClearType setting as I do. Otherwise I would have spotted it. Well done.
Joined: 11/22/2004
Posts: 1468
Location: Rotterdam, The Netherlands
Bisqwit wrote:
No, for the purpose of shortening URLs one should create a shortened URL scheme, spefically linking "this corresponds to this".
Not to knock specifically shortened URLs, but what I mean is that URLs should never be too long, even when you've got plenty of room. This tends to occur in particular with blogs that put the entire title in simplified form in the URL: they're just really long and difficult to read. It's hard to make nice and succinct URLs, though. If we were to use your example (/TAS/SMB1byKlmz) we'd have to simplify/abbreviate the titles of the games, manually make sure we're using consistent abbreviations, and it might be confusing if someone's nickname starts with a lowercase character. But besides, even if we ever radically change the database, we could always just copy the ID column and rename it "unique ID", then use that for lookups rather than the actual primary key.
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
la mammal wrote:
Wow. Your gif worked perfectly on me. You must use the exact same ClearType setting as I do. Otherwise I would have spotted it. Well done.
Windows XP default or something. But didn't you spot it? Otherwise you would not have posted.
NitroGenesis
He/Him
Editor, Experienced player (556)
Joined: 12/24/2009
Posts: 1873
Too bad it can be easily spotted in quotes.
YoungJ1997lol wrote:
Normally i would say Yes, but thennI thought "its not the same hack" so ill stick with meh.
Editor, Experienced player (570)
Joined: 11/8/2010
Posts: 4036
I only spotted it because I saw something move out of the corner of my eye while I was reading the rest.
Joined: 2/26/2011
Posts: 98
CoolKirby wrote:
I only spotted it because I saw something move out of the corner of my eye while I was reading the rest.
That's how I noticed it. It's a fun idea. I've done that myself on another message board.