Hi, And sorry for the delay. Attila Lendvai skribis: >> > It's not enough to just handle the thrown exception, because we need to be >> > able to get hold of the fetched content, too. >> >> Would it make sense, then, to use the lower-level ‘http-get’ from (web >> client)? That would let the code deal with all the HTTP idiosyncrasies. > > > i think it boils down to this trade-off: > > 1) keep http-fetch simpler, at the expense of reimplementing parts of > it in the go importer (e.g. the redirection logic) > > 2) add this extra complexity to http-fetch, and avoid the extra > complexity of a local, potentially half-assed %http-fetch in the go > importer. > > 3) something else i'm not aware of For now, I’m somewhat in favor of #1. My take would be: try to implement whatever’s needed specifically for the Go importer; from there, we can eventually revisit that situation and maybe switch to something that’s more like #2. How does that sound? Thanks, Ludo’.