Is anyone *actually* using 2.7 ? Because my testing shows it's been kinda borked for a while and I have had a rather insufficient amount of bug reports for if anyone actually is.

I am just wondering why anyone would want to keep using it, for this particular purpose? Running gogrepoc on a Windows XP PC or some very old Linux on their home PC that doesn't have support for python 3.x?

EDIT: Well, even XP is supported at least by some versions/branches of python 3... Does gogrepoc require certain version of python 3, or is e.g. python 3.4.4 enough for it (which according to google works on XP, at least)?

I also take back what I said about Python 2 handling things better than Python 3 for the serial codes, the print function and interactive shell certainly do (they do some kind of implicit conversion magic ) but as soon as you try and write to file and read again it's worse than Python 3 ( the implicit conversion magic simply odes not happen when you're writing to a file so you just write a mess).

I used to have legacy python applications that did not work with python 3, and since I did not want to have two python installations on my computer, for several years I used gogrepoc with python 2.7. Now that I no longer used those legacy programs, I have moved to python 3.

Oh Python 2.7 I had forgotten how much I hate you, why does shutil.rmtree have completely different (and far worse) longpath handling on Windows than anything in os or file IO ? Also why do none of it's errors indicate that the problem is the path is too long and instead say the file can't be accessed ?

Going to change the behaviour of verify just to make things more consistent with the general philosophy:

The default behaviour for handling files that are questionable / in the wrong location should be to orphan them (unless there's strong reasons to do otherwise eg it's fine to delete incomplete downloads that are no longer in the manifest since these are just garbage ). The option to delete instead should generally exist ( but it's not super high priority since trash exists ), and the option to leave in place is only really relevant for verify ( it's generally infeasible in most places because the usual reason to *zap* files is to replace them except for clean and if you want files left in place *why are you running clean ?* literally the only thing it does is *zap* files ) .

So verify will now orphan files that fail the checks by default with the option not to provided.

Okay, got everything except the archival set verification done now, that doesn't need to be done online, so I'm going to put the rest of this through it's paces, I might upload without the archival set test to Dev when that's done , since I'm running so late with this update.

Normal update worked fine, so time for the full update.

This should also let me see if there is still unflagged GOG stuff since I haven't done a full update since the middle of Jan and I have no outstanding new/updated games now so anything that gets caught by this is either update or has an inconsistent last modified across servers.

Anyone know what's going on with downlink ? Have they changed it ? I'm getting 403's from all downlink reads and visiting the link in the browser shows error messages so it's not just a client thing.

It may have been a temporary thing.

I've just gotten manifest updates and downloaded files listed in them for the following games, without issue.

Blackwell Legacy|1207662883
Colt Canyon|1964712512
Crystal Project Demo|1548183851
Death Trash Demo|1606404506
Heroine's Quest: The Herald of Ragnarok|1237516203

But then, I am using gogcli.exe not gorepo.py, though that probably shouldn't make any difference.

Are the /downlink/ links even official? I have never seen them referenced on GOG. The only official links are the /downloads/ links. I have just checked myself. While the redirection works for the /downlink/ links the CDN throws an error message instead of sending you the file. The /downloads/ links on the other hand work as expected.

It seems GOG has recently made more changes to the CDN than just forbidding the multiple slashes...
Nope, it still doesn't work. Using gogcli can be the difference if gogcli uses the official /downloads/ links.

I only use downlink as a fallback but GOG forgets the MD5 data occasionally so every time I do a full update I hit downlink like 20 times to see if the MD5 data is there ( it almost never is )

What do you mean by that? That you tried to access an .xml file but couldn't because of an error? I just made a complete rescan of my database with my own downloader and all xml files were accessible on the first try. And this has been consistently reliable recently.

I mean in the case where the MD5 data is missing from the primary site (download) it's pretty much always also missing on downlink.

Yes, it's much improved, missing MD5 data these days is mainly support packages for Mac (like the Language packs for BGIII ), so if you don't have a fairly large collection and download Mac versions you won't run into it, I think in my like 2400 games, there's only 1 or 2 Windows games, I'll have to make a list again.* There's also a tar.gz that's marked as an installer (some of those do have MD5s.

*Will probably be next month since I have to turn off all the MD5 skipping for installers.

In terms of games that were updated by weren't marked

brigand_oxacaca ( though I think this might be a case of me catching the initial update pass on GOG ( I can see there's a download available now which is probably the other two OSs )
Broken sword shadows of the templars reforged
Crossed Swords

Which is a huge improvement over previously where full updates would pull down an extra 100-300GB of updates.


Also I have a *lot* of extra with changed Last-Modified but I think that's an artifact since the vast majority of them have the same date, so it's probably just a new set of servers or a sync that happened rather than new data.