Sven, thank you so much for the detailed explanation. I used it to look at the debugger again so I could reformulate the way I look for bugs. I’d was looking backwards from the error, but had I started at the DoIt, I would have seen the ZnBivalentWriteStream and seen the class comment. Moreover, I would have realized it all starts with the Transcript. I guess habits from my other environment (iOS, where the stack leads into closed source). And here in Pharo I can inspect the MethodContext and see the program counter :)
Thanks again, -Edward > On Dec 29, 2014, at 11:16 AM, Sven Van Caekenberghe <s...@stfx.eu> wrote: > > Hi Edward, > >> On 29 Dec 2014, at 17:19, Edward Povazan <empova...@gmail.com> wrote: >> >> I am playing around with Zinc, and I get an >> "Error - Improper store into indexable object” >> at >> ZnBufferedWriteStream>>nextPut: >> >> My Workspace: >> ----- >> | response | >> response := ZnClient new >> url:'http://amazon.com'; >> get; >> response. >> >> response writeOn: Transcript. >> Transcript flush >> ----- >> >> The reason is that instead of a Character, there is a SmallInteger of value >> 10 being added at ByteString>>at:put: (and 10 is a linefeed). >> >> Also the string in the entity is a WideString >> (ZnUTF8Encoder>>next:putAll:startingAt:toStream is where the choice is made >> to decode a ByteString vs WideString). >> >> Is this a bug? Running on Pharo 3. My newb debugging skills don’t go further >> :( Other sites that fail are reddit.com, news.ycombinator.com. >> >> Thanks, >> -Ed > > Short answer: > > Zinc writes to or reads from binary streams, the Transcript is a character > stream. I know that what you are trying to do is actually in the > documentation and in the API, and it normally works, just not in all cases. > It is better to use inspection to look at what is going on, you'll get a much > more structured (object-oriented) view of what is going on. I would suggest > inspecting the ZnClient instance, which will hold the state of the last > request and response. > > Long answer: > > If you try > > | response | > response := ZnClient new > url:'http://stfx.eu/small.html'; > get; > response. > > response writeToTranscript. > > It will work and print > > HTTP/1.1 200 OK > X-Pad: avoid browser bug > Date: Mon, 29 Dec 2014 18:56:37 GMT > Content-Length: 113 > Etag: "202f9-71-4c7ae94364f1e" > Server: Apache/2.2.22 (Ubuntu) > Accept-Ranges: bytes > Last-Modified: Mon, 20 Aug 2012 08:48:51 GMT > Vary: Accept-Encoding > Content-Type: text/html > > <html> > <head><title>Small</title></head> > <body><h1>Small</h1><p>This is a small HTML document</p></body> > </html> > > So what is the difference ? > > The fact that the responses that you get from the website you mention are > using chunked transfer encoding > (http://en.wikipedia.org/wiki/Chunked_transfer_encoding). Zinc deals with > this in both ways, while reading a response, but also while writing a > response (which is what you are trying to do here). > > Here is how you disable chunked transfer encoding so that you can write the > response to the Transcript > > | response | > response := ZnClient new > url:'http://amazon.com'; > get; > response. > > response headers removeKey: 'Transfer-Encoding'. > response writeToTranscript. > > Zinc manages to write to the Transcript in most cases by cheating a bit: it > wraps a ZnBivalentWriteStream around another stream when needed - such a > stream can do both binary and character writing at the same time (though the > result is only good for quick and dirty debugging, since it is not 100% > correct). > > I hope the long explanation didn't make it worse ;-) > > HTH, > > Sven > > PS: BTW, in Pharo 4 with Zinc #bleedingEdge you get even nicer and more high > level presentations of Zinc objects in GT-Inspector ! > >
