Weblogs for simonw
#429
Posted by simonw on Wed 28 Sep 2011 at 23:22
I looked in the Error Console and saw "XStringBundle (too much recursion)" which immediately took me to:
http://getsatisfaction.com/mozilla_messaging/topics/3_1_2_does_not_work
Which said "disable Mail Tweaks" so I did and all my folders reappeared.
http://getsatisfaction.com/mozilla_messaging/topics/3_1_2_does_not_work
Which said "disable Mail Tweaks" so I did and all my folders reappeared.
[0 Comments
| Add Comment
|
]
]
#428
Posted by simonw on Mon 5 Sep 2011 at 16:31
Oh joy - as an aid to helping my employers sell ".UK" domain name Nominet have decided what people really need is a prize draw. So they emailed me several times from a third party provider to tell me this on all our accounts email addresses, suggesting we give people a way to opt in on the domain registration form.
No doubt next week we can look forward to a Nominet Lottery, if we aren't buried in spam telling me I've won the Nominet prize draw just download this PDF/EXE/FLV file.
What would help sell ".UK" domain name would be a registry who are so pedantic on registration details I generally advise people not to promote their ".co.uk" names but stick with ".com" and ".net" which will carry on working even if they make some silly omission in telling them of some change of circumstance.
No doubt next week we can look forward to a Nominet Lottery, if we aren't buried in spam telling me I've won the Nominet prize draw just download this PDF/EXE/FLV file.
What would help sell ".UK" domain name would be a registry who are so pedantic on registration details I generally advise people not to promote their ".co.uk" names but stick with ".com" and ".net" which will carry on working even if they make some silly omission in telling them of some change of circumstance.
[0 Comments
| Add Comment
|
]
]
#427
Posted by simonw on Tue 23 Aug 2011 at 11:12
Under KDE4 with KDE and update-alternatives browser set correctly Icedove insisted on opening URLs in my copy of Firefox 6 in ~/Desktop/firefox rather than Google-Chrome in /usr/bin/google-chrome.
After some searching around and much use of grep found the culprits in .gconf/desktop/gnome/url-handlers
Removing ".gconf" and restarting Icedove and it now behaves correctly under KDE.
This took too much finding, I didn't expect Icedove to use GNOME settings on KDE.
After some searching around and much use of grep found the culprits in .gconf/desktop/gnome/url-handlers
Removing ".gconf" and restarting Icedove and it now behaves correctly under KDE.
This took too much finding, I didn't expect Icedove to use GNOME settings on KDE.
[0 Comments
| Add Comment
|
]
]
#426
Posted by simonw on Mon 1 Aug 2011 at 17:38
My employers built one of (the?) first online website builders.
This led in a very early IT bubble to the formation of ZY PLC, and resulted in the corporate history of this small limited company having an unusually complex diagram, which ties it in vaguely with many of the early UK Internet companies, and the founding of Nominet. Although somehow they managed to avoid becoming fashionably wealthy (at least that is what they tell me when discussing pay).
I joined after all this fancy stuff happened, and think we should have sold the domain a long time ago (probably one of the intervening bubbles would have been best but it is easy in hindsight).
Having been an early ISP ZYNET (the companies whose assets it inherited and many of it's clients!) have a number of attractive domain names.
Anyway when SEDO get their act together we should have a link on http://zy.com/ for buying the "zy.com" domain if you feel the need.
This led in a very early IT bubble to the formation of ZY PLC, and resulted in the corporate history of this small limited company having an unusually complex diagram, which ties it in vaguely with many of the early UK Internet companies, and the founding of Nominet. Although somehow they managed to avoid becoming fashionably wealthy (at least that is what they tell me when discussing pay).
I joined after all this fancy stuff happened, and think we should have sold the domain a long time ago (probably one of the intervening bubbles would have been best but it is easy in hindsight).
Having been an early ISP ZYNET (the companies whose assets it inherited and many of it's clients!) have a number of attractive domain names.
Anyway when SEDO get their act together we should have a link on http://zy.com/ for buying the "zy.com" domain if you feel the need.
[0 Comments
| Add Comment
|
]
]
#425
Posted by simonw on Sun 3 Jul 2011 at 20:30
Another coaster.
Colour me unimpressed, I expect DELL shipped a LOT of drives just like mine.
Colour me unimpressed, I expect DELL shipped a LOT of drives just like mine.
[0 Comments
| Add Comment
|
]
]
#424
Posted by simonw on Fri 1 Jul 2011 at 21:55
Pondered why Icedove (based on Thunderbird 3.0.11) doesn't always seem to do a stable sort.
I mean this is a fixed bug, right?
On closer inspection "sort by from/recipient" sorts by more than the displayed information (sender/recipient name, or email if this is not given), it always includes their email address for sorting purposes.
Thus I was expecting my emails to be sorted by "From" and then "Date" but that correspondent routinely uses many email addresses (not least they often email me when they have a problem with an email address).
Search Messages by using a general search for text in the "From" field does the right thing easily enough if you realise why it has "failed".
I'm sure there is a bug here, but I'm not sure I know precisely what it is or what to do about it. Possibly the "From" & "Recipient" fields should include proper email details greyed out (the trend to hide email addresses in email clients and other "technical" information is a perpetual frustration of my life). Probably Apple's mail client would have colour coded the different recipients in it's glorious multicolour way, which whilst it induces some sort of guttural revulsion in me would have visually indicated that these recipients were deemed to be different from each other (the same thing could be done with two colours or other visual clue, using a hint only if the displayed value is the same but the sorted value is different).
My own software rarely gets such scrutiny, but then the user base is a lot smaller, although the lack of such scrutiny could be one of the reasons for that.
I mean this is a fixed bug, right?
On closer inspection "sort by from/recipient" sorts by more than the displayed information (sender/recipient name, or email if this is not given), it always includes their email address for sorting purposes.
Thus I was expecting my emails to be sorted by "From" and then "Date" but that correspondent routinely uses many email addresses (not least they often email me when they have a problem with an email address).
Search Messages by using a general search for text in the "From" field does the right thing easily enough if you realise why it has "failed".
I'm sure there is a bug here, but I'm not sure I know precisely what it is or what to do about it. Possibly the "From" & "Recipient" fields should include proper email details greyed out (the trend to hide email addresses in email clients and other "technical" information is a perpetual frustration of my life). Probably Apple's mail client would have colour coded the different recipients in it's glorious multicolour way, which whilst it induces some sort of guttural revulsion in me would have visually indicated that these recipients were deemed to be different from each other (the same thing could be done with two colours or other visual clue, using a hint only if the displayed value is the same but the sorted value is different).
My own software rarely gets such scrutiny, but then the user base is a lot smaller, although the lack of such scrutiny could be one of the reasons for that.
[0 Comments
| Add Comment
|
]
]
#423
Posted by simonw on Wed 15 Jun 2011 at 17:35
As per bug 134702, I share his pain....
I would like to make a directory portable within and between web servers, allowing copy, and rename, and not break Apache authentication. This falls down because AuthUserFile is absolute or relative to server root. I think I'd like it to be relative to htaccess as per patch, or perhaps embedded inline in the htaccess file as that would be even less breakable.
Guess I have my patch, and Apache Auth is not widely used (but that is largely a chicken and egg situation. Add "Logout" to a few browsers and pretty up the interface and I'm sure more people would use it).
Obviously my directory contains .htaccess and .htpasswd and could (would) end up in the published space, but as noted elsewhere every servers excludes ".ht*" files for precisely this reasons. And I'm less worried about encrypted passwords being cracked than the basic usability.
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=134702
Is there a tidy solution to my issue? Or such I just patch the module and move on?
I would like to make a directory portable within and between web servers, allowing copy, and rename, and not break Apache authentication. This falls down because AuthUserFile is absolute or relative to server root. I think I'd like it to be relative to htaccess as per patch, or perhaps embedded inline in the htaccess file as that would be even less breakable.
Guess I have my patch, and Apache Auth is not widely used (but that is largely a chicken and egg situation. Add "Logout" to a few browsers and pretty up the interface and I'm sure more people would use it).
Obviously my directory contains .htaccess and .htpasswd and could (would) end up in the published space, but as noted elsewhere every servers excludes ".ht*" files for precisely this reasons. And I'm less worried about encrypted passwords being cracked than the basic usability.
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=134702
Is there a tidy solution to my issue? Or such I just patch the module and move on?
[0 Comments
| Add Comment
|
]
]
#422
Posted by simonw on Tue 14 Jun 2011 at 18:37
It started as looking at an issue with running wkhtmltopdf on Squeeze and poor quality text.
Noted that if I print a PDF from a web page using the "Print to File" option on my KDE desktop using Debian Squeeze I get the same issue (or visually similar issue suggesting the issue is lower in the stack than wkhtmltopdf) in the PDF.
Convinced myself it isn't entirely a kerning issue in Evince (which seems to be the best PDF viewer available in Squeeze) by printing the PDF and still showing the same issue in the printed copy (so after pdftops presumably).
This bug seemed relevant but the problem isn't THAT bad, and affects content more than titles. But it could be another manifestation of the same bug.
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=556867
"Kate" seems to print boring old text to PDF just fine on the few cases I've tested but some pages from Firefox and Chrome seem to just space the letters in an ugly fashion - the ugliness is similar in both browsers although it isn't always the same place on the web page in each browser that go ugly, but then there are slight difference in the rendering and printing so not surprising spacing isn't identical (although it is pretty close between the browsers).
Anyone any ideas on this before I raise a bug, or comment on the existing one? I could do with a refresher of "Print to File". I presume this uses the cups-pdf driver which use to require you to install a specific PPD. But presumably the issue lies either in a lower layer yet (Cairo) or in how that lower level in invoked. Can I inspect fonts to get an overview of hinting information, as I understand the behviour can change with resolution (does this apply to creating PDFs?).
Hard to give specific examples as I think somethings are rendering issues.
This URL works fine:
https://developer.mozilla.org/samples/cssref/font-family.html
This one the text in the Google Ad I got mangled (Words "Document Management" the "g" and "e" are too close together when viewed in evince at 200%, but viewed at other zoom levels the text is fine. So is that the same issue, or is that an evince rendering issue? There are so many layer affecting all this that it is hard to rip them away with confidence (unless I guess you are use to working in this field).
http://www.loadpdf.com/ebook/debian-guide.html
Noted that if I print a PDF from a web page using the "Print to File" option on my KDE desktop using Debian Squeeze I get the same issue (or visually similar issue suggesting the issue is lower in the stack than wkhtmltopdf) in the PDF.
Convinced myself it isn't entirely a kerning issue in Evince (which seems to be the best PDF viewer available in Squeeze) by printing the PDF and still showing the same issue in the printed copy (so after pdftops presumably).
This bug seemed relevant but the problem isn't THAT bad, and affects content more than titles. But it could be another manifestation of the same bug.
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=556867
"Kate" seems to print boring old text to PDF just fine on the few cases I've tested but some pages from Firefox and Chrome seem to just space the letters in an ugly fashion - the ugliness is similar in both browsers although it isn't always the same place on the web page in each browser that go ugly, but then there are slight difference in the rendering and printing so not surprising spacing isn't identical (although it is pretty close between the browsers).
Anyone any ideas on this before I raise a bug, or comment on the existing one? I could do with a refresher of "Print to File". I presume this uses the cups-pdf driver which use to require you to install a specific PPD. But presumably the issue lies either in a lower layer yet (Cairo) or in how that lower level in invoked. Can I inspect fonts to get an overview of hinting information, as I understand the behviour can change with resolution (does this apply to creating PDFs?).
Hard to give specific examples as I think somethings are rendering issues.
This URL works fine:
https://developer.mozilla.org/samples/cssref/font-family.html
This one the text in the Google Ad I got mangled (Words "Document Management" the "g" and "e" are too close together when viewed in evince at 200%, but viewed at other zoom levels the text is fine. So is that the same issue, or is that an evince rendering issue? There are so many layer affecting all this that it is hard to rip them away with confidence (unless I guess you are use to working in this field).
http://www.loadpdf.com/ebook/debian-guide.html
#421
Posted by simonw on Tue 7 Jun 2011 at 10:50
Noted that the project wkhtmltopdf works well for those people that want to build a system to do this, rather than just use a tool. A couple of the online services to do same are clearly using this tool under the hood.
We currently do in-house web page thumbnailing with a Python script that uses gtkmozembed which I've been meaning to share properly since I've "improved" it over the copy I found lying around somewhere (it even has options). Since it doesn't require a modified Qt library the Python script is probably still preferable for the purposes of quick and dirty web page thumbnailing tool than wkhtmltoimage, but since the options are similar I can see both happening.
Anyone else done similar? There is a PHP class for wrapping wkhtmltopdf in the repository but I suspect we'll write some Perl, as that is what we usually do.
I've already shared "PDF: Unfit for Human Consumption" with my colleagues, but we aren't all human.
http://www.useit.com/alertbox/20030714.html
We currently do in-house web page thumbnailing with a Python script that uses gtkmozembed which I've been meaning to share properly since I've "improved" it over the copy I found lying around somewhere (it even has options). Since it doesn't require a modified Qt library the Python script is probably still preferable for the purposes of quick and dirty web page thumbnailing tool than wkhtmltoimage, but since the options are similar I can see both happening.
Anyone else done similar? There is a PHP class for wrapping wkhtmltopdf in the repository but I suspect we'll write some Perl, as that is what we usually do.
I've already shared "PDF: Unfit for Human Consumption" with my colleagues, but we aren't all human.
http://www.useit.com/alertbox/20030714.html
[0 Comments
| Add Comment
|
]
]
#420
Posted by simonw on Fri 3 Jun 2011 at 14:10
Comment 15 on the Ubuntu bug did the trick
dmesg was saying:
58611.309469] scsi 7:0:0:0: Direct-Access 1.01 PQ: 0 ANSI: 0
[58611.310768] sd 7:0:0:0: Attached scsi generic sg4 type 0
[58611.315486] sd 7:0:0:0: [sdd] 93249 512-byte logical blocks: (47.7 MB/45.5 MiB)
[58611.322565] sd 7:0:0:0: [sdd] Write Protect is off
[58611.322576] sd 7:0:0:0: [sdd] Mode Sense: 03 00 00 00
[58611.322583] sd 7:0:0:0: [sdd] Assuming drive cache: write through
[58611.336457] sd 7:0:0:0: [sdd] Assuming drive cache: write through
[58611.336466] sdd: sdd1
[58611.358510] sd 7:0:0:0: [sdd] Assuming drive cache: write through
[58611.358519] sd 7:0:0:0: [sdd] Attached SCSI removable disk
[58611.398453] sd 7:0:0:0: [sdd] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[58611.398461] sd 7:0:0:0: [sdd] Sense Key : Illegal Request [current]
[58611.398468] sd 7:0:0:0: [sdd] Add. Sense: Logical block address out of range
[58611.398476] sd 7:0:0:0: [sdd] CDB: Read(10): 28 00 00 01 6c 40 00 00 01 00
[58611.398492] end_request: I/O error, dev sdd, sector 93248
The disk identified itself as:
lrwxrwxrwx 1 root root 9 Jun 3 14:06 usb-__123456789ABCDEF-0:0 -> ../../sdd
lrwxrwxrwx 1 root root 10 Jun 3 14:06 usb-__123456789ABCDEF-0:0-part1 -> ../../sdd1
However after the DD suggested in comment 15 below GNOME opened a window when the camera was reconnected.
https://bugs.launchpad.net/ubuntu/+source/util-linux/+bug/500398/comments/15
dmesg was saying:
58611.309469] scsi 7:0:0:0: Direct-Access 1.01 PQ: 0 ANSI: 0
[58611.310768] sd 7:0:0:0: Attached scsi generic sg4 type 0
[58611.315486] sd 7:0:0:0: [sdd] 93249 512-byte logical blocks: (47.7 MB/45.5 MiB)
[58611.322565] sd 7:0:0:0: [sdd] Write Protect is off
[58611.322576] sd 7:0:0:0: [sdd] Mode Sense: 03 00 00 00
[58611.322583] sd 7:0:0:0: [sdd] Assuming drive cache: write through
[58611.336457] sd 7:0:0:0: [sdd] Assuming drive cache: write through
[58611.336466] sdd: sdd1
[58611.358510] sd 7:0:0:0: [sdd] Assuming drive cache: write through
[58611.358519] sd 7:0:0:0: [sdd] Attached SCSI removable disk
[58611.398453] sd 7:0:0:0: [sdd] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[58611.398461] sd 7:0:0:0: [sdd] Sense Key : Illegal Request [current]
[58611.398468] sd 7:0:0:0: [sdd] Add. Sense: Logical block address out of range
[58611.398476] sd 7:0:0:0: [sdd] CDB: Read(10): 28 00 00 01 6c 40 00 00 01 00
[58611.398492] end_request: I/O error, dev sdd, sector 93248
The disk identified itself as:
lrwxrwxrwx 1 root root 9 Jun 3 14:06 usb-__123456789ABCDEF-0:0 -> ../../sdd
lrwxrwxrwx 1 root root 10 Jun 3 14:06 usb-__123456789ABCDEF-0:0-part1 -> ../../sdd1
However after the DD suggested in comment 15 below GNOME opened a window when the camera was reconnected.
https://bugs.launchpad.net/ubuntu/+source/util-linux/+bug/500398/comments/15
[0 Comments
| Add Comment
|
]
]