While Vasquez's work doesn't rule out a meteor, and in fact he specifically states that the weather was not exceptional, I don't think it was a meteor because that doesn't seem to be consistent with the times of the ACARS messages. Personally I think lightening and turbulance together are the most likely causes as listed there.
Given the exterior temperature, I dont think icing was a factor. However, something was off on the airspeed indicators and this might have been the start of the problem.
Hmm.... I actually think they might have been going too fast.
Consider that there are approx. 4 minutes of ACARS warning messages transmitted before the crash. The first of these suggest that the computers were getting bad input data and handing full manual control back to the pilots. Given current reports of ACARS messages, I don't think the pilots knew what their airspeed was. My own thinking is that maybe a lightening strike damaged airspeed sensors, either physically or electronically.
High altitudes pose a number of specific problems in that the performance margin between stall speed (too slow) and the point where aerodynamic stability is lost (too fast) is quite small. When you add turbulance to this (which causes fluctuations in the plane's airspeed), this can get very tricky.
Also consider that the plane would likely have been downed by turbulance while exiting the storm (see the meterological analysis at weathergraphics.com). This does not suggest an updraft or downdraft was proximally responsible.
Here is my proposed timeline:
0200Z, pilot enters main multicell convective system. 0209Z, lightening strike damages external airspeed sensors, depending on where the airplane was struck, might have caused EMI regarding comminications between components as well (EMI events have caused problems on other airbus airplanes). 0210Z, ACARS: alternate law engaged (i.e. safety features disabled), autopilot disabled.
My guess at this point the pilot may not have had a solid idea of of airspeed, and was thus forced to fly the plane under dangerous conditions without vital information. My guess further is that the plane was going faster than recommended and that lack of stability was causing additional probems when combined with turbulance.
0210-0213Z, ACARS: multiple failures of instruments. Possibly caused by physical damage by severe turbulance.
0213Z, ACARS: full flight system failure 0214Z, ACARS: vertical cabin speed
Presumably the plane entered into a spin around the last ACARS message and broke apart in the air shortly after the last ACARS message.
Personally, I think a meteor collision is unlikely because the failures occur over 10 minutes. I would expect a meteor collision to occur when the other ACARS errors were not sent. I think the ACARS messages seem to suggest somthing other than a meteor.
One possibility is that a lightening strike could have physically or electronically damaged some sort of sensor, and that this could have caused a failure cascade to the computer systems. This would explain the disengangement of the autopilot and at that point, the pilot's job becomes very difficult due to near-maximum speed, turbulance, and high altitude. In short the performance margin becomes very narrow. Additionally, once alternate law is enguaged the problems become more severe.
If this is the case, it would seem to me possible that the plane was flying in conditions that even experienced pilots could not fly manually though with any safety, and that flying along the edge of the performance margin could have lead to more instability which could have lead to physical damage to the control systems. Eventually the plane loses control and breaks up due to aerodynamic stress.
Anyway, that is my working theory. It suggests that the fatal lightening strike would have occurred prior to the FIRST of the troubling ACARS messages and at least 10 minutes before the crash.
The invisibility of the titles depends on which version of the story you're looking at. Apparently they have some versions that are cached or something like that. On my system, if I go to the slashdot homepage and then click on one of the "Read more..." links, the titles of comments are white on a white background, so they're invisible.
Interesting. On my system only seems to affect pages from science.slashdot.com.
I wonder if the issue is a missing background image.
When we forked LedgerSMB from SQL-Ledger, the project was pretty much as you describe.
My suggestion: 1) Analyze the db. 2) Start building a NEW application which is properly structured and engineered (rather than merely written quickly) and as this happens use headers to redirect to/from the old application. 3) Eventually retire the old application.
This may take a year or even two. However, in the end it will save your employer a ton of money, headaches, and outages.......
Let's write off IE6 as obsolete and force those users to upgrade.
That is the approach we take with LedgerSMB. The market share of IE 6 and even 7 is not enough to justify supporting it as a browser when Firefox can be installed side-by-side, and when the cost to doing so is horrible i18n hacks.
Well, the optimal way would be setting up with ipsec (shouldn;t be hard to do) and krb5 (harder, but doable). krb5 then allows you to do passwordless authentication and session encryption and you could use ipsec to provide an additional layer of (probably stronger) encrpytion the ip payload level.
Particularly when using Krb5 for session encryption;-) Of course how many slashdotters know that you can use telnet with session encryption on an intranet? Slim to none?
Ummm.... In printing, tables aren't necessarily tabular:-) That is one of the big mismatches.
XHMTL table -> Latex tabular and friends XHMTL floating div -> LaTeX table, figure, etc.
They are not at all the same things. XML and HTML folks tend to use "table" to mean "tabular data" while it has a very specific layout meaning in the world of print which has NOTHING to do with tabular data.
On 1: What do you mean by separation between table/figure floats? In CSS, can't you just assign two different classes that are both floated, but have other different properties? There's some fundamentally different kind of float that you need?
Ok. In print a "table" is one kind of float. A "figure" is another. A table of figures generally only includes the former. I suppose you could create classes of divs in this sort of way, but then generating tables of figures would be problematic, for example. I didnt see anything in Prince's engine which handled this gracefully.
On 2: Not that I'm discounting the possibility, but Prince's layout engine isn't good enough for you? What's wrong with it?
It didn't look like 2-page layouts properly handled gutter margins from their samples. Proper margin management on double-sided bound documents is a key feature of LaTeX.
On a third level, crossreference, index, etc. management didn't look graceful using PrinceXML. Also, none of their examples had footnotes, so I couldn't judge those. Being able to put a footnote which says "See also page x below" where x is the actual page where information is found on is quite helpful, and having the pagenumber autogenerate (and even hyperlink in the PDF) to its target is something I use all the time.
I also keep wondering how Word will "replace LaTeX" as a commonly used "editor." I don't know ANYONE who uses LaTeX as an editor.... I don't even know if it can be used this way.
I don't know about the rest of the LaTeX folk here but my editor is VIM. LaTeX is my typesetting engine.
You have a fundamental issue here which is that most of the print media css stuff I have come across are:
1) Inflexible (hard to tweak layout on the occasion where it needs to be tweaked, no separation between table and figure floats, etc-- both are fundamental issues with SGML) 2) Layout depends on your conversion engine, and I haven't seen any extremely good engines of this sort that use css directly.
HTML/CML + CSS gives you a subset of what LaTeX can give you.
The problem though is you can't disable the styles and only focus on semantic markup. It IS possible to create structured documents in Word, but it takes practice and discipline, and the temptation is there to do your layout at the same time you do your content.
From a publishing perspective, this is a bad thing because you can't count on the state of submitted content.
I can guarantee you they are not. They are a small publisher targetting a niche market (and they don't even sell through bookstores). And mostly they publish the owner's books (and a few of his associates' works).
When I published my book, I did it all in LaTeX. It saved me between $500 and $1000 in setup fees and the output looks awesome! I had a few issues with hyphenations and line-breaking, but those were fairly minor and fixed during the proofreading phase. Extra money I could pay for editorial help..... Heck, after playing around with a few less able packages (Scribus, etc), I did the cover design in LaTeX. Took a little time but not too bad.
One warning though.... Once you go down the road and design a book, you will never look at any book you read again. First you will look at the cover and the cover design. Then you will look at the page design, font layout, kerning, spacing, etc. Then you will actually read the book. It will become an obsession.
My experience with book design is one of the few things that prevents me from creating a truly free OCR-B font (I don't want to be obsessed with font details).
Re:Much more than you think leaves Word & Co.
on
MS Word 2010 Takes On TeX
·
· Score: 2, Interesting
Actually I know of one publisher which uses Word from start to finish. And the quality of typesetting is sadly evident. Fortunately they have the sense usually to use a ragged right edge.
The more I work with LaTeX on books, the happier I am with it. It really is a good program.
When I am writing a one-off business letter, word processors are the way to go.
However, if I need a document that will be maintained, needs to be formatted in a consistent way, etc. there is NO WAY I am touching a word processor for that job.
If I am writing a very complex document that needs maintenance, usually I do it in LaTeX as a source tree with svn etc.
I wrote my book (http://books.google.com/books?printsec=frontcover&id=GkQeAhqaybUC) using LaTeX. It was much easier, given the thorns, eths, etc. than trying to constantly use the character map and Word. Also I found I could do some really cool things in LaTeX that were extremely difficult in other environments. Some things were harder (diagrams), but that is what xfig is for:-).
In general, I think for serious authoring and typesetting, nothing can touch LaTeX. It is extremely powerful, robust, and full-featured. Moreso in all of these areas than any other package I have ever worked with.
This being said..... 80-90% of the advanced stuff done in my book was pulled out into macros in a custom.sty. This was so that the book would be mostly plain text, and the.sty would handle the layout. All in all, it was a very, very good experience.
Word -> OpenOffice-> Docbook -> LaTeX, provided you have a decent template in Word......
The basic thing is that usually what you want from a pubishing perspective is a nice, structurally defined file. You want to eventually get this into a format like LaTeX where you can do the actual typesetting.
After writing my book in LaTeX ("The Serpent and the Eagle: An Introduction to the Elder Runic Tradition") I have decided I am extremely happy with LaTeX's layout and typesetting abilities. However these elements are done last. If you have a decent tool chain, you can import from Word, OpenOffice, DocBook, etc.
This being said, DocBook is not as structurally rich as LaTeX (no separation of tables and floats in LaTeX terminology, for example-- DocBook tables are all LaTeX tabular or similar) so something is lost.
Slashdot Mirror
While Vasquez's work doesn't rule out a meteor, and in fact he specifically states that the weather was not exceptional, I don't think it was a meteor because that doesn't seem to be consistent with the times of the ACARS messages. Personally I think lightening and turbulance together are the most likely causes as listed there.
Given the exterior temperature, I dont think icing was a factor. However, something was off on the airspeed indicators and this might have been the start of the problem.
Hmm.... I actually think they might have been going too fast.
Consider that there are approx. 4 minutes of ACARS warning messages transmitted before the crash. The first of these suggest that the computers were getting bad input data and handing full manual control back to the pilots. Given current reports of ACARS messages, I don't think the pilots knew what their airspeed was. My own thinking is that maybe a lightening strike damaged airspeed sensors, either physically or electronically.
High altitudes pose a number of specific problems in that the performance margin between stall speed (too slow) and the point where aerodynamic stability is lost (too fast) is quite small. When you add turbulance to this (which causes fluctuations in the plane's airspeed), this can get very tricky.
Also consider that the plane would likely have been downed by turbulance while exiting the storm (see the meterological analysis at weathergraphics.com). This does not suggest an updraft or downdraft was proximally responsible.
Here is my proposed timeline:
0200Z, pilot enters main multicell convective system.
0209Z, lightening strike damages external airspeed sensors, depending on where the airplane was struck, might have caused EMI regarding comminications between components as well (EMI events have caused problems on other airbus airplanes).
0210Z, ACARS: alternate law engaged (i.e. safety features disabled), autopilot disabled.
My guess at this point the pilot may not have had a solid idea of of airspeed, and was thus forced to fly the plane under dangerous conditions without vital information. My guess further is that the plane was going faster than recommended and that lack of stability was causing additional probems when combined with turbulance.
0210-0213Z, ACARS: multiple failures of instruments. Possibly caused by physical damage by severe turbulance.
0213Z, ACARS: full flight system failure
0214Z, ACARS: vertical cabin speed
Presumably the plane entered into a spin around the last ACARS message and broke apart in the air shortly after the last ACARS message.
Personally, I think a meteor collision is unlikely because the failures occur over 10 minutes. I would expect a meteor collision to occur when the other ACARS errors were not sent. I think the ACARS messages seem to suggest somthing other than a meteor.
One possibility is that a lightening strike could have physically or electronically damaged some sort of sensor, and that this could have caused a failure cascade to the computer systems. This would explain the disengangement of the autopilot and at that point, the pilot's job becomes very difficult due to near-maximum speed, turbulance, and high altitude. In short the performance margin becomes very narrow. Additionally, once alternate law is enguaged the problems become more severe.
If this is the case, it would seem to me possible that the plane was flying in conditions that even experienced pilots could not fly manually though with any safety, and that flying along the edge of the performance margin could have lead to more instability which could have lead to physical damage to the control systems. Eventually the plane loses control and breaks up due to aerodynamic stress.
Anyway, that is my working theory. It suggests that the fatal lightening strike would have occurred prior to the FIRST of the troubling ACARS messages and at least 10 minutes before the crash.
The invisibility of the titles depends on which version of the story you're looking at. Apparently they have some versions that are cached or something like that. On my system, if I go to the slashdot homepage and then click on one of the "Read more..." links, the titles of comments are white on a white background, so they're invisible.
Interesting. On my system only seems to affect pages from science.slashdot.com.
I wonder if the issue is a missing background image.
titles of comments are invisible? Not on my system (ff3, fedora 10, etc).
What system are you running?
When we forked LedgerSMB from SQL-Ledger, the project was pretty much as you describe.
My suggestion:
1) Analyze the db.
2) Start building a NEW application which is properly structured and engineered (rather than merely written quickly) and as this happens use headers to redirect to/from the old application.
3) Eventually retire the old application.
This may take a year or even two. However, in the end it will save your employer a ton of money, headaches, and outages.......
But not always w3m.
Let's write off IE6 as obsolete and force those users to upgrade.
That is the approach we take with LedgerSMB. The market share of IE 6 and even 7 is not enough to justify supporting it as a browser when Firefox can be installed side-by-side, and when the cost to doing so is horrible i18n hacks.
Well, the optimal way would be setting up with ipsec (shouldn;t be hard to do) and krb5 (harder, but doable). krb5 then allows you to do passwordless authentication and session encryption and you could use ipsec to provide an additional layer of (probably stronger) encrpytion the ip payload level.
Particularly when using Krb5 for session encryption ;-) Of course how many slashdotters know that you can use telnet with session encryption on an intranet? Slim to none?
Ummm.... :-) That is one of the big mismatches.
In printing, tables aren't necessarily tabular
XHMTL table -> Latex tabular and friends
XHMTL floating div -> LaTeX table, figure, etc.
They are not at all the same things. XML and HTML folks tend to use "table" to mean "tabular data" while it has a very specific layout meaning in the world of print which has NOTHING to do with tabular data.
On 1: What do you mean by separation between table/figure floats? In CSS, can't you just assign two different classes that are both floated, but have other different properties? There's some fundamentally different kind of float that you need?
Ok. In print a "table" is one kind of float. A "figure" is another. A table of figures generally only includes the former. I suppose you could create classes of divs in this sort of way, but then generating tables of figures would be problematic, for example. I didnt see anything in Prince's engine which handled this gracefully.
On 2: Not that I'm discounting the possibility, but Prince's layout engine isn't good enough for you? What's wrong with it?
It didn't look like 2-page layouts properly handled gutter margins from their samples. Proper margin management on double-sided bound documents is a key feature of LaTeX.
On a third level, crossreference, index, etc. management didn't look graceful using PrinceXML. Also, none of their examples had footnotes, so I couldn't judge those. Being able to put a footnote which says "See also page x below" where x is the actual page where information is found on is quite helpful, and having the pagenumber autogenerate (and even hyperlink in the PDF) to its target is something I use all the time.
I also keep wondering how Word will "replace LaTeX" as a commonly used "editor." I don't know ANYONE who uses LaTeX as an editor.... I don't even know if it can be used this way.
I don't know about the rest of the LaTeX folk here but my editor is VIM. LaTeX is my typesetting engine.
Not quite.
You have a fundamental issue here which is that most of the print media css stuff I have come across are:
1) Inflexible (hard to tweak layout on the occasion where it needs to be tweaked, no separation between table and figure floats, etc-- both are fundamental issues with SGML)
2) Layout depends on your conversion engine, and I haven't seen any extremely good engines of this sort that use css directly.
HTML/CML + CSS gives you a subset of what LaTeX can give you.
The problem though is you can't disable the styles and only focus on semantic markup. It IS possible to create structured documents in Word, but it takes practice and discipline, and the temptation is there to do your layout at the same time you do your content.
From a publishing perspective, this is a bad thing because you can't count on the state of submitted content.
Also, if the publisher wants to change the layout, they can change the .sty afterwards.....
I can guarantee you they are not. They are a small publisher targetting a niche market (and they don't even sell through bookstores). And mostly they publish the owner's books (and a few of his associates' works).
When I published my book, I did it all in LaTeX. It saved me between $500 and $1000 in setup fees and the output looks awesome! I had a few issues with hyphenations and line-breaking, but those were fairly minor and fixed during the proofreading phase. Extra money I could pay for editorial help..... Heck, after playing around with a few less able packages (Scribus, etc), I did the cover design in LaTeX. Took a little time but not too bad.
One warning though.... Once you go down the road and design a book, you will never look at any book you read again. First you will look at the cover and the cover design. Then you will look at the page design, font layout, kerning, spacing, etc. Then you will actually read the book. It will become an obsession.
My experience with book design is one of the few things that prevents me from creating a truly free OCR-B font (I don't want to be obsessed with font details).
If you want to see some good output from LaTeX:
http://books.google.com/books?printsec=frontcover&id=GkQeAhqaybUC#PPA38,M1
Actually I know of one publisher which uses Word from start to finish. And the quality of typesetting is sadly evident. Fortunately they have the sense usually to use a ragged right edge.
The more I work with LaTeX on books, the happier I am with it. It really is a good program.
I think you hit the nail on the head.
When I am writing a one-off business letter, word processors are the way to go.
However, if I need a document that will be maintained, needs to be formatted in a consistent way, etc. there is NO WAY I am touching a word processor for that job.
If I am writing a very complex document that needs maintenance, usually I do it in LaTeX as a source tree with svn etc.
I wrote my book (http://books.google.com/books?printsec=frontcover&id=GkQeAhqaybUC) using LaTeX. It was much easier, given the thorns, eths, etc. than trying to constantly use the character map and Word. Also I found I could do some really cool things in LaTeX that were extremely difficult in other environments. Some things were harder (diagrams), but that is what xfig is for :-).
In general, I think for serious authoring and typesetting, nothing can touch LaTeX. It is extremely powerful, robust, and full-featured. Moreso in all of these areas than any other package I have ever worked with.
This being said..... 80-90% of the advanced stuff done in my book was pulled out into macros in a custom .sty. This was so that the book would be mostly plain text, and the .sty would handle the layout. All in all, it was a very, very good experience.
Actually XML + CSS can already do most of these things. Just not to the point of professional typesetting :-)
Word -> OpenOffice-> Docbook -> LaTeX, provided you have a decent template in Word......
The basic thing is that usually what you want from a pubishing perspective is a nice, structurally defined file. You want to eventually get this into a format like LaTeX where you can do the actual typesetting.
After writing my book in LaTeX ("The Serpent and the Eagle: An Introduction to the Elder Runic Tradition") I have decided I am extremely happy with LaTeX's layout and typesetting abilities. However these elements are done last. If you have a decent tool chain, you can import from Word, OpenOffice, DocBook, etc.
This being said, DocBook is not as structurally rich as LaTeX (no separation of tables and floats in LaTeX terminology, for example-- DocBook tables are all LaTeX tabular or similar) so something is lost.