Browser misinterprets relative link! 1

[robert89]

Hi there,
We use relative linking where possible and recently have started to see some of these links show up in our failure report.
The link as it is on our page:
../../../../english/exhibits/art_comp/pics/irving_gordon.jpg

the link as it shows up in the failure report
/french/exhibits/english/exhibits/art_comp/pics/irving_gordon.jpg

This interpretation of the link does not occur with IE, Netscape, or Opera.

Wondering if someone knows something about other browsers that might be reinterpreting.

Info/thoughts greatly appreciated.
Bob

 

[PennyB]

What's the url of your page?

 

[bcastner]

Could you describe whether it occurs under the same OS version?

I think you are running into an issue where the OS is unable to parse the directory depth you are passing. Each version of Windows has different limits on how deep a directory path can be.

I think if you shortened the path by creating less deeply sourced Folders and adjusting the code you would have success.

 

[robert89]

hi bcastner,
Like those thoughts. Do you know what the maximum directory depth can be when all versions of windows are considered. I am assuming by OS you mean the OS of the visitor, which of course, I have no way of connecting the OS to the filepath being misinterpreted. This appears to be a recent problem.

Perhaps, it's Microsofts latest xo!)($$## etc. OS.

But thanks for the direction of thought. I will pursue.

Bob

 

[bcastner]

The hard thing is you would have to count the characters for the expanded string, including the fact that it will be expanded using long file names for everything, and not short names.

I do not have a ready reference, but any NT based OS such as Win2k/XP can handle up to 255 characters (the last is reserved), and the Win9x I believe substantially fewer.

I wish I had a reference for you, but I have seen the same thing from batch files, where the expansion would work in Win2k and fail in Win98, and offer it as one suggestion.

My second, related, thought, is that the NT family of OS versions are more comfortable with essentially POSIX syntactical statements (your example is one). The earlier non-NT OS versions have no clue about this directory pathing.

 

[robert89]

Hi bcastner,
Thanks for the info. I will do some research based on these toughts and if I come up with something, will post.

Thanks again,
Bob

 

[robert89]

Hi bcastner,
If you catch this, I have a supplementary question which my be related.
We use Dreamweaver, and graphic files, jpg only, keep being removed from their various directories on our test server. Not all, but some. There one day, gone the next. Some of these files, I haven't actually paid enough attention to determine if it is all files, have filenames ending in a number.

Wondering, if it is possible that they are deleted because their reference within our documents may be longer than OS allotment, and or, are their some systems where numbers at the end of a file name may interfer?

Thanks,
Bob

 

[bcastner]

robert89,

Do you mean removed as in gone from the folders, or they work one day and not the next but physicly remain in the folder?

 

[robert89]

Yep, removed, gone, never to be seen again. When we discover the missing files, we pulled them back from the live server to the test server. Asked our IT guy to look into it and he has, but can't determine any reason why this might be happening. We, the web guys, are preplexed.

Thanks,
Bob

 

[bcastner]

I know it is malware, but I am not sure what. Do the steps outlined in faq608-4650

It is not Dreamweaver or IIs removing those files.

 

[bcastner]

robert89,

Please post back with your results. This is a genuinly odd problem and I am curious as to how it sorts for you.

Bill