bug in uri decoding

Poster	Content
jrogers nk4um User Posts: 101	<div>Thanks Tony,<br/><br/>I have a workaround for the immediate problem that brought this to my attention, so I don''t need a fix right now, but a proper solution would be appreciated soon.<br/><br/>Incidentally, my immediate problem was caused by using the javascript builtin function encodeURIComponent which generates escape sequences using lowercase hex digits. The workaround is to use escape() instead, however that is not expected to work with non-ascii characters (although I have not tested this)<br/><br/>I think this escaping bug is also responsible for a behavior that I noticed where certain uris can be called as subrequests but not externally via the http transport.<br/><br/>-J<br/><br/>PS: I''d like to add that bugs like this are why I love open source software. I was able to track the problem down through the source, isolate it, and understand it; and if I needed to I could have fixed it myself. Without the source I would have been stuck scratching my head for a LOT longer.</div> 2009-11-06T18:15:05.000ZNovember 6, 2009 18:15 Thanks Tony, I have a workaround for the immediate problem that brought this to my attention, so I don''t need a fix right now, but a proper solution would be appreciated soon. Incidentally, my immediate problem was caused by using the javascript builtin function encodeURIComponent which generates escape sequences using lowercase hex digits. The workaround is to use escape() instead, however that is not expected to work with non-ascii characters (although I have not tested this) I think this escaping bug is also responsible for a behavior that I noticed where certain uris can be called as subrequests but not externally via the http transport. -J PS: I''d like to add that bugs like this are why I love open source software. I was able to track the problem down through the source, isolate it, and understand it; and if I needed to I could have fixed it myself. Without the source I would have been stuck scratching my head for a LOT longer.
tab nk4um Moderator Posts: 485	<div>Great catch Jeff! <br/><br/>We''ll get this fixed. What timescale to you need the fix for NK3 in?<br/><br/>Cheers,<br/>Tony</div> 2009-11-06T12:28:14.000ZNovember 6, 2009 12:28 Great catch Jeff! We''ll get this fixed. What timescale to you need the fix for NK3 in? Cheers, Tony
jrogers nk4um User Posts: 101	<div>Not sure what you guys are using for bug tracking these days, but I just tripped across a pretty serious bug in uri decoding. I filed it at <br/><br/>http://bugxter.netkernel.org/bug/22<br/><br/>The fix isn''t complicated, but it does affect a pretty low-level component (layer1 in nk3, layer0 in nk4).</div> 2009-11-06T05:03:35.000ZNovember 6, 2009 05:03bug in uri decoding Not sure what you guys are using for bug tracking these days, but I just tripped across a pretty serious bug in uri decoding. I filed it at http://bugxter.netkernel.org/bug/22 The fix isn''t complicated, but it does affect a pretty low-level component (layer1 in nk3, layer0 in nk4).

jrogers

nk4um User

Posts: 101

<div>Thanks Tony, I have a workaround for the immediate problem that brought this to my attention, so I don''t need a fix right now, but a proper solution would be appreciated soon. Incidentally, my immediate problem was caused by using the javascript builtin function encodeURIComponent which generates escape sequences using lowercase hex digits.  The workaround is to use escape() instead, however that is not expected to work with non-ascii characters (although I have not tested this) I think this escaping bug is also responsible for a behavior that I noticed where certain uris can be called as subrequests but not externally via the http transport. -J PS: I''d like to add that bugs like this are why I love open source software.  I was able to track the problem down through the source, isolate it, and understand it; and if I needed to I could have fixed it myself.  Without the source I would have been stuck scratching my head for a LOT longer.</div>

November 6, 2009 18:15

Thanks Tony,

I have a workaround for the immediate problem that brought this to my attention, so I don''t need a fix right now, but a proper solution would be appreciated soon.

Incidentally, my immediate problem was caused by using the javascript builtin function encodeURIComponent which generates escape sequences using lowercase hex digits. The workaround is to use escape() instead, however that is not expected to work with non-ascii characters (although I have not tested this)

I think this escaping bug is also responsible for a behavior that I noticed where certain uris can be called as subrequests but not externally via the http transport.

-J

PS: I''d like to add that bugs like this are why I love open source software. I was able to track the problem down through the source, isolate it, and understand it; and if I needed to I could have fixed it myself. Without the source I would have been stuck scratching my head for a LOT longer.

tab

nk4um Moderator

Posts: 485

November 6, 2009 12:28

Great catch Jeff!

We''ll get this fixed. What timescale to you need the fix for NK3 in?

Cheers,
Tony

jrogers

nk4um User

Posts: 101

November 6, 2009 05:03bug in uri decoding

Not sure what you guys are using for bug tracking these days, but I just tripped across a pretty serious bug in uri decoding. I filed it at

http://bugxter.netkernel.org/bug/22

The fix isn''t complicated, but it does affect a pretty low-level component (layer1 in nk3, layer0 in nk4).