[TYPO3-core] RFC #13750: Bug: Several GIFBUILDER features broken for IM4, IM6 and GM

[Joerg Wagner [DigiLog]]

I now checked through the complete documentation of the ImageMagick 
command-line tool. The parameters used in Georg's testcase are 
definitely wrong. An integer in square brackets following a file name is 
allowed for source files and selects a single frame in a multi frame 
image format, but must be enclosed INTO the quotes and may never be 
appended AFTER them! (See section "Selecting Frames" of 
http://www.imagemagick.org/www/command-line-processing.html)

So all file parameters in the testcase that have appended brackets must 
be changed from...
	"C:/Users/...fe_maskNT.gif"[0]
to
	"C:/Users/...fe_maskNT.gif[0]"

With the latter form all testcases should work with my patches.

Cheers,
Jörg.


schrieb Joerg Wagner [DigiLog] am 20.05.2010 09:41:
> Hi Steffen,
>
> you have used the unit test cases posted by Georg Grossberger in bug
> report 13750. As I stated in my note in that report, the function will
> never pass these tests (neither with nor without the patch) as the
> function was never designed to handle parameter content that is
> partially quoted but has appended parts OUTSIDE the quotes, e.g.:
> "C:/Users/Someuser.Domain/Documents/Htdocs/typo3temp/temp/61401f5c16c63d58e1d92e8a2449f2fe_maskNT.gif"[0]
>
>
> IMHO such parameters are a direct contradiction to the meaning of the
> term "quoted". Here is what I wrote to Georg (see notes in the bug report):
>
> ----- snip -----
> Considering your testcases:
> You write that many tests failed with the patched version. The original
> function version (with or without my patch) will indeed fail on all of
> your testcases which have a closing quote that is NOT at the end of the
> quoted string. The function expects the opening and closing quotes to
> enclose the WHOLE string. But many of your cases have an appended [0]
> after the quoted part of one of the parameters. Does this really happen?
> I am not much aware of the IM and GM parameter structure, but this looks
> like a contradiction to me. Maybe I am overlooking something here but I
> would argue that you cannot quote a parameter and then put a part of it
> outside of the quotes. What is this parameter structure...
> "path/filename.gif"[0]
> ...meant for that you use so often in your testcases?
>
> If these appended [0] parts in your testcases should indeed be
> irrelevant, than I strongly suggest to stay with the original function
> and my patch. All the remaining testcases should work well with it.
> ----- snap -----
>
> Cheers, Jörg.
>
>
> schrieb Steffen Gebert am 10.05.2010 12:30:
>> Am 10.05.2010, 10:24 Uhr, schrieb JoH asenau <info at cybercraft.de>:
>>
>>> Come on folks, this is a major bug, that *breaks important IM/GM
>>> features
>>> for any Windows user*, so where is the problem about comitting this
>>> patch? -
>>> Instead of concentrating on new features for upcoming T3 versions,
>>> IMHO we
>>> should fix those major bugs and blockers of the current versions first.
>>
>> Hi Joey and Jörg,
>>
>> I'm willing to review this, unfortunately I don't have good luck with
>> the version Jörg sent on Tuesday.
>>
>> As the IM commands are a very sensible part of TYPO3, I'd really like to
>> see unit tests for the changed function(s). For me (on MacOS) only the
>> first 2 tests work - even with the patch.
>> Nevertheless, I attached the testcase file applied against typo3_src
>> (and the same for the trunk-patch, which is also CGL-conform now).
>>
>> What I don't understand is, how this should break IM4. I tried Susanne's
>> snippet under windows, but nothing was broken with IM4. Looking at the
>> code in t3lib_div::imageMagickCommand(), I can see that
>> unQuoteFilenames() is only called with $switchCompositeParameters =
>> TRUE.
>> But this is only the case for gm and im6, isn't it?
>>
>> By setting it TRUE by hand, I was able to observe the different behavior
>> of unQuoteFilenames(), which indeed looks weird, but seems not to harm
>> in my case (because I use IM4?).
>>
>> Could we agree to add the unit tests first and then fix this?
>>
>> If you can give me a hint, how this should affect IM4, I'd also be happy
>> ;-)
>>
>> Steffen