SEP
24
2001

Konqueror and Other Version 5 Browsers

Peter-Paul Koch,
maintainer of a JavaScript/DOM/CSS testing site, recently updated
his browser
section
. In it he reviews
Konqueror, together with
the other Version 5 browsers, Mozilla
and IE.
He concludes, "In short, the few remaining bugs in Konqueror are
details that no doubt will be solved soon. The development team has
succeeded in building an excellent, standards compliant browser from
scratch. Therefore I expect Konqueror to become a real competitor for
Netscape 6 on Linux: it has made an excellent start and can only
become better
."

Comments

Look at http://www.w3.org/Style/CSS/ That page is 100% html and css compliant and so far konqueror is the only browser I can find that actually properly renders that page. Mozilla 0.9.4, opera 5, netscape 4.77, ie 6, etc all fail. Mozilla gets it closest behind konqueror but I think I could read war and peace in the time it takes to scroll the page. Opera scrolls quickly and has the floating menu but it doesn't look even close to correct. IE doesn't have the menu float and has lots of other rendering issues. Netscape 4 .... well let's just not discuss that one.

From doing a lot of web development and extensively using CSS I have found nothing that is even close to konqueror in terms of CSS support. Too many times I find browsers that just can't get stuff right. I would rate Mozilla and Opera and pretty good CSS support. Not excellent but still pretty good and mozilla has been showing a lot of improvements. I don't doubt it will be excellent in another 6 months or so. IE6 I find to be pretty poor in CSS support.

Overall I find konqueror to be my favorite browser to use and love using it for website devel. I can make stuff 100% spec and verify that it works well. Then I just make the changes needed for each of the various browsers needed to make them vew it properly also and stick it in a browser specific version server side and register them with the right browsers. That way all the browsers get the right page and it is far less painful to develop the stuff.


By kosh at Mon, 2001/09/24 - 5:00am

I can't see the floating menu at all (KDE 2.2.1 RHJ 7.1 Ben's RPMs) - Mozilla and Opera bot show it correctly (Opera is pretty fast in scrolling, too). Besides, the layout in Konq. is probably a bit incorrect as well ('CSS Browsers' and 'Authoring Tools' are on the same line - should be slightly offset IMO).

- Biswa.


By Biswapesh at Mon, 2001/09/24 - 5:00am

I have been looking at it with KDE 2.2.1 and I see a nice transparent blue menu that floats and scrolls quickly down the page. It is fully transparent through the entire page. With Opera here it is just completely transparent while it should be this blueish alpha blended menu. Opera does scroll quickly but it doesn't look very good.

I am not sure what the alignment of those two items should be. I was looking at the CSS for the page and it looks fairly complex. I couldn't say exactly where those two items should be. On my system they don't look like they are on the same line but they are fairly close.

If I flick the scroll wheel down on Mozilla 0.9.4 to scroll the page on my k6-2 500 it will take about 10 minuts to reach the bottom and that is not an exaggeration. I have no idea why that page is so slow on mozilla on this box. Every other page I have hit in mozilla seems to be pretty fast. 768M of ram, 2.4.10 kernel and a 64MB DDR Radeon card.

This is all a debian sid system using the kde packages that are in sid.


By kosh at Mon, 2001/09/24 - 5:00am

> can't see the floating menu at all (KDE 2.2.1 RHJ 7.1 Ben's RPMs)
> ('CSS Browsers' and 'Authoring Tools' are on the same
> line - should be slightly offset IMO).

No such problems here (KDE 2.2 on SuSE 7.2).

Greetings,
Tackat


By Torsten Rahn at Mon, 2001/09/24 - 5:00am

> I can't see the floating menu at all (KDE 2.2.1 RHJ 7.1 Ben's RPMs) -
> Mozilla and Opera bot show it correctly (Opera is pretty fast in scrolling, > too).

No problem with KDE 2.2.1 (kdelibs-2.2-22mdk) on Linux-Mandrake 8.0.
Menu is present, transparent! :-)

> Besides, the layout in Konq. is probably a bit incorrect as well ('CSS
> Browsers' and 'Authoring Tools' are on the same line - should be slightly
> offset IMO).

Try to enlarge browser's window to full-screen.
I am on 1024x768, and still it's not enough (small overlap of 'CSS Browsers' and 'Authoring Tools' text)... Probably W3C designed this page for 1280x1024 resolution.

As about CSS support in Konqueror - pls check
http://www.konqueror.org/content/khtml_css2.html
It has a lot of very useful info.

Cheers,

Vadim


By Vadim Plessky at Tue, 2001/09/25 - 5:00am

On my KDE2.2.1 the menu is there, semi-transparent. Mozilla is unbearably slow on this page (as kosh wrote, it really takes minutes to scroll down the page). And CSS Browsers and Authoring tools are offset by about 1/4 of letter height.

Hey, I just love konqueror. :-))))))


By Lada at Wed, 2001/09/26 - 5:00am

Yes, Konqueror has great basic and semi-advanced CSS support. Using Konqueror as my primary test environment, I've developed a few web pages with heavy CSS formatting. I've never had to do much tweaking to make it look nice in other browsers.

However, Konqueror fails on the really advanced CSS aspects. Check out the CSS tests at http://www.bath.ac.uk/~py8ieh/internet/home.html (especially the 'Evil test suite').


By Gaute Hvoslef K... at Mon, 2001/09/24 - 5:00am

Admittedly, Mozilla sucks at the W3C site (hopefully they fix this soon). However, overall CSS support is still probably better in Mozilla than Konqueror.

For example, check this cool CSS demo site:

http://www.meyerweb.com/eric/css/edge/complexspiral/demo.html

There are some cosmetic problems with Konqueror, such as with the menu on the left. (Also note that IE can't handle the page correctly either.)


By Joni K. at Mon, 2001/09/24 - 5:00am

This page looks fine to me using the Konqueror in the HEAD branch. What problems are you refering too?

Rich.


By Richard Moore at Mon, 2001/09/24 - 5:00am

The only cosmetic issue with this page under KDE 2.2.1 is that the mouse-overs for the menu on the left don't work 100% correctly - sometimes there are horizontal bands of the previous colour left there.

I haven't got KDE HEAD up and running on my machine yet, so I can't confirm whether it's gone away or not.


By Jon at Tue, 2001/09/25 - 5:00am

Can Konqueror style form elements now? I'm using 2.1.1 now, and all the widgets shows up as they are specified in KDE theme, not as in CSS rules. IMO it's a long way to claim CSS compliance if buttons and text boxes cannot be styled using CSS. Along with that, KHTML 2.1.1 has tons of rendering bugs concerning CSS. Has situation been improved?


By Danny at Tue, 2001/09/25 - 5:00am

Color etc information are suggestions to the browser and not a requirement. It is within spec to ignore that stuff. Actually I would like it to be configurable. Personally I hate websites that try and change the look and color of buttons, textareas etc since I have kde setup to look like how I want. So what we need is a checkbox in kde that is something along the lines of allow all site/ sites in this list/ no sites to override the look and feel of the standard controls.

Every point release of kde since 1.0 has offered major improvements each time in html/css rendering capabilities. 2.2.1 is far superior to 2.1.1 and fixed lots of things.


By kosh at Tue, 2001/09/25 - 5:00am

Yes, those "color etc information" are suggestions, and browsers are under no obligation to render as specified. But ignoring them does not lead to the claim of CSS compliance.

Certainly some people pefer the form widgets to be consistent with their desktop, but how is a web author to know how the user desktops are set up (or can the user change their desktop settings at all)? Normally the web author would write different style sheets, one for regular viewing, and the other for visually-impaired (high-contrast, large font). Without the ability to style form widgets, the high-contrast style sheet may not be very useful. (side question: does Konqueror 2.2 allow users to select alternate style sheet, like in Mozilla?)

I've seen a lot of improvements in KHTML since the original 2.0 release, but for advanced features like CSS 2, DOM 2, etc., it still seems very lacking compared to Mozilla and IE. Nevertheless, I'd say Konqueror surpasses Opera and certainly Netscape 4, so congratulations to the KDE team.


By Danny at Tue, 2001/09/25 - 5:00am

>side question: does Konqueror 2.2 allow users to select alternate style sheet, like in Mozilla?

Yes. It has a nice wizard that makes you a custom stylesheet.


By not me at Tue, 2001/09/25 - 5:00am

You're refering to user-defined style sheet which applies to every web the user visits. This is different from "alternate stylesheet" which specify an alternative representation of a web page.

It's specified in the HTML by:

Under Mozilla, you can select such style sheet by View->Use Stylesheet->foo. I can't find anyway to do this in other major browsers.


By Danny at Fri, 2001/09/28 - 5:00am

Konqueror has been able to style form elements for some time, but only if you are using a KDE widget style that adapts to specified colours - the default highcolor style does for instance, as do all the Qt styles, and Mosfet's Liquid style.

If you're using a pixmap-based theme (maybe you're using the Acqua or Acqua-Graphite styles from themes.org, or the Marble or System styles, or one of Vadim Plessky's styles), then KStyle cannot reliably change the colours of widgets, as they are all pre-drawn by a human rather than created by code that can adapt to different colours.

Switch back to the default highcolour style and you will see Konqueror does indeed style form elements.


By Amazed of London at Tue, 2001/09/25 - 5:00am

Yeah, sure, here is Konqueror reading a simple xml file:

Connection HealthClient connection information.1Connection Usage1Connection Usage01001000[Threads_connected] [max_connections]Usage10001Traffic2Traffic01201024001^[Bytes_sent] 102400Usage10001Number of SQL Queries3Number of SQL Queries0110101^[Com_select] 1010001Memory HealthGives an overview of critical memory health issues.2Query Cache Hitrate1Query Cache Hitrate01001000 (^[Qcache_hits]/(^[Qcache_hits]+^[QCache_inserts] +^[QCache_not_cached]))*100100Hitrate10001Key Efficiency2Key buffer size02301000[Key_blocks_used]*[key_cache_block_size] [key_buffer_size]Key buffer usageKey buffer10001Key buffer usage01001000100-(^[Key_reads]/^[Key_read_requests])*100100Hitrate10002

Why? No default on style sheet, so garbage comes out. The page is mysqladmin_health.xml from MySQL, which also seems to have this "simple answers" problem spreading throughout the Internet and programming world.

KDE and Konqueror are suffering from the same synaptic collapse; it won't work and you can't fix it with a tool that requires it to be working in the first place!

Kong could use a searchable help file, and content. MySQL could use some cyanide to slow it down a little in its overdevelopment. If you can't start at the beginning, "Why doesn't xml render in Konqueror?" then there is no sense in going into a long lecture of how great it is.

I think both programs are great, but they could both use some professional manual writers and editors, not programmers, real manual writers.

Not even css has the best css support; there is no css support, period.

Everything is on "runaway" development. Would anyone care to tell how to make Konqueror render an xml file? Without it being embedded in an html file! What's the point of having files inside of files inside of files in an object oriented system? If Konqueror can't find a style sheet, it should ask what to do and suggest a style sheet with proper instructions on how to get the job done.

And this is exactly what's lacking in all the patting themselves on the back development that is going too fast. Too fast causes crashes; witness the most popular OS and its millions, literally, of problems.

Send me an email.


By CyberSongs at Mon, 2004/12/06 - 6:00am

How on earth would you render "an xml file"? XML is meta-container format and specifies nothing about rendering or even format.
Konqueror doesn't have problems with XHTML though.


By carewolf at Mon, 2004/12/06 - 6:00am

Render it by processing it. Either as PHP, which is what most xml's are currently, or perl, perhaps. All basic php's are seen as xml in supplied mysql program provided scripts. They cannot be displayed without being converted to html of some sort. The files in phpMyAdmin, for example, begin with "<?php" and that should be a flag for at least two alternatives, one if mysql and/or php is installed to render by either, or two, if neither is installed and/or have problems, to alternatively display the code with a message that php/mysql did not handle the page.

But if the browser thinks that "<?php" file type xml, then it should do a better job and not ask for a plugin, or worse, display a blank page, then the browser is wrong in its definitions.

Phrases like "meta container format" are long grammatical statements equivalent to "it's an eXtended Markup Language file." Regardless, it should not result in a blank page.

For example, having a simple file called packets01.xml, Konqueror will ask for a plugin while IE will display the page in colored code. This file was created by Ethereal. It can be exported as various file types. The idea is to make the file viewable in a tabular format such as php, but the Ethereal authors used type .xml

There are programs that can open such files, but Konqueror will ask for them and selection of the wrong file will produce useless rendering.

Saved as a PostScript document, the file can be read using GhostView, but it would be preferable to see it in tabular form using php.

I believe you thought I had made a simple statement. There was all of this and more in that statement. Nothing comes down to such a simple aside as it's this kind of file or that kind of file that can't be rendered. All files can be rendered, it doesn't matter what kind they are, and there are choices in how to have them rendered. Rendering means making it human friendly viewable, nothing more. Hard definitions of words like "render" do not exist in programming as all definitions are variable in programming.

Succesful development is always based on what the end user wants, not what the developer thinks he should give them.

I'm currently debugging the MySQL source code line 1830, etc., for socket failures inherent in MySQL. Care to comment?

Ciao


By CyberSongs at Wed, 2004/12/08 - 6:00am

If all you want is to see the tag soup, open it in kate.

It will highlight all the brackets.

How that is supposed to be useful, I have no idea.

XML contains nop presentation info, unless it's a known XML variety.

For example:

* fwbuilder saves firewall configs as XML.
* kword saves documents as XML
* I save arbitrary python object trees as XML

Most of those can't be shown "in tabular form". And all are XML.

In short: I think you are very confused.

And about the MySql stuff.. are you somehow mislead into believing this is a MySql support forum?


By Roberto Alsina at Wed, 2004/12/08 - 6:00am

Where does this attitude come from? Why is it that whenever anyone posts a comment, there's always someone out there waiting to pounce on them? Roberto, I really don't think you're old enough to decide whether or not someone is confused. Pretty sure of that.

Why personal? Why can't you simply discuss the subject?

I'll bet I was working in Unix before you went to school.

Don't flame me; I have no time for such nonsense.

By the way, I was creating style sheets when html came out.

I had suggested a windows type interface for unix and linux probably before you knew what the Internet was.

Therefore, since I was associated with KDE's beginnings, back when Dragonlord was in Dragonsrealm on dal.net, I think I know a little about kde's development.

Where were you when I was working on some original GeOS windows, and before that on graphics for the pdp11 and 8088?

Where were you when I was authoring "relocate" in machine and assembly code?

And MySQL is increasingly becoming part and parcel of an html based database; where were you when I wrote papers about the proposed html database structure, or suggested it to Tim Berners-Lee?

And for that matter, what are the cloud chamber photographs of pi-mesons and other subatomic particles?

Where were you when I was working with and studying particle physics?

Don't "aside" me again with such childish flaming techniques.


By CyberSongs at Wed, 2004/12/08 - 6:00am

He is not trying to flame you, but just explain to you that you are seriously confused. For instance PHP is a server-side script, that cannot be processed by a browser. If a webserver delivers PHP or MySQL to the browser it has been misconfigured.

The only issue where you might be right is the XML. I think we put in code that treats all code delivered as XML as XHTML which will mean an empty page if it is not. We could propably do some more fine-grained detection here.


By carewolf at Wed, 2004/12/08 - 6:00am

Well, if what you say makes no sense, you are probably confused, and age has nothing to do with it.

I seriously doubt you were hacking UNIX before I went to school, unless you worked on AT&T labs, since I went to school in 1974.

As for where I was when you were writing relocate in assembly, well, I was probably old enough to know that writing it in C was a better idea (although I have no idea what relocate is suppsed to be).

Dude, I *have* programmed Fortran on a PDP-11. I have spent two years studying differentiation theory, and all that has nohing whatsoever to do with this. You can be a hotshot physicist, and still have no idea what XML is. XML doesn´t imply presentation, unless you know what variant it is, and that´s it. End of argument.

On the other hand, if you believe that your personal qualifications (or mine) have anything to do with the argument, you are, let´s say, stupid.


By Roberto Alsina at Wed, 2004/12/08 - 6:00am

I'm as stupid as Albert Einstein.

Relocate is the machine code that everyone now uses to make all modules dynamic, that is, they are loaded to a relative address, the relocate routine adjusts the base of the execution stack for the code to the new relative address.

All "jumps" and any uncondtional goto or branch is replaced by a conditional relative branch.

XML is problematic. KDE and X are problematic. Konqueror is problematic. They all seem to be reflective of a Microsoft gui style of programming. They don't know how to manage either the screen refresh or memory.

Maybe I was stupid in asking a question about XML. I didn't think so. Apparently, however, it left an open door for someone to call me stupid. Being able to see through transparent people, I knew instinctively that I'd be flamed for asking any question in public. However, I was more brave than stupid, so I asked. I felt like taking the position of the average person that comes along and asks a simple question. How many of them are made to feel bad by so-called IT professionals by being monikered with "stupid?" It is just not right to call someone who asks a question "stupid."

Sorry, it is just not right.

If you wouldn't do it to a 5 year old, don't do it to an adult. And if you do that to a 5 year old, you need professional help. In fact, I think if you do it to an adult you need professional help.

I find it hard to believe that a question about XML's validity would raise hecklers from the dead. I don't much care about XML. We all use it, but it's not all that life-sustaining to defend it to the death. Sometimes it works, and sometimes it doesn't. CSS and style sheets are going to come and go; to be replaced, eventually, by something not so dependant upon dependancies. And Konqueror will only improve; as will KDE and Linux. None of them are important enough to offend other people by calling their questions and opinions "stupid." That's just not right, and, it's mean.

For KDE, Konqueror, and the rest of the so-called super-gui's:

I wrote another routine, years ago, that uses one execution clock to change a screen. I recently suggested KDE look into how it's done. I had an old Commodore refreshing the screens faster than any current OS.

If people would pay attention, they might learn something.

You don't transfer the 3 meg or so of bits to the monitor [televsion, regardless of its current name], you change the pointer to the display area. Doing that, you can change the screen, call it refresh if you like, in one microprocessor execution clock.

Given a sufficient area in memory, you can have as many screens as you like available instantly. Currently, it's my bet that all of these gui programs stream the new screen to the display area; that is inefficient in a world of pointers, object oriented code, and Vector Operators, which are all really Indirect Reference Pointers and calls for arrays, binary tables, etc..

If the L1 or L2 cache has the necessary depth, of course, if not, it has to come from external memory or disk memory. Even with the external RAM onboard though, it is much faster than gooeying the refresh from a stream. If the display chipset, on the other hand, had known this technique, there would be no hashing for local memory or disk memory.

The X, KDE, and Konqueror I'm using right now are eating up to 30% cpu time doing absolutely nothing. And because it shows up as a delay between screens, as in using ALT+TAB, like Windows does, I know it is the display code mechanisms. In fact, it looks so much like a Microsoft programmer wrote it, I begin to wonder.

I just happened to have noticed that newer programs are slowing down in performance. I've seen this many times before on mainframes, years ago.

And if I said I was not more than ten feet from J.Presper Eckert when Unix was intially discussed, I meant it, even if said in passing. Unix and Aix were oldies, but goodies. Both ended in the Roman Numeral IX because of the 9 companies involved; not just Bell Labs!

Bell Labs takes far too much credit, especially when the people I worked with were supplying them and AT&T with the machines and the network on which to test UNI6 and Unix.

But then again, let's not forget that when Alexander Graham Bell's future father-in-law was going to patent the telephone on the morrow, Bell and his fiancee beat daddy to it by going to the patent office with daddy's drawings on that day!

I suppose you weren't aware that Alex Bell had stolen the plans and design of the telephone from his future father-in-law.

It's in the history books, if you're looking for a reference.

Sometimes some people do know a little more and at least, as Solomon once said, you would be wiser to listen first.

KDE and Konqueror are great, and I like them, but they definitely still need work.

If they implement the pointer screen change as stated here, no gui or browser will be faster. I'm surprised the video chipset manufacturers haven't standardized this yet. Perhaps they should. It would give them something to work on with high profit potential.

Oh, btw, I and my colleague back in the year you stated at Bell Labs, built a radio and fried a resistor. The interesting thing was that it turned into a diode. So, we figured, that's how Wang discovered it too. By accident, the first solid state diode, and hence junction, is discovered, and somebody gets the Nobel Prize for it. Then goes on to burn two resistors and put them opposite polarity end-to-end and discover, further, that they can control the current throught the thing at the junction. Hence, the first transistor. This is meant to be funny. I remember laughing quite good when we did it and saying "So that's how he discovered solid state electronics!"

Lastly, most things at Bell Labs do happen by accident, as in
"Watson, come here, I need you" Alexander Graham Bell while producing the first working telephone, er, by accident.

Everything is not so high-fallutin as the world would make us believe. The greatest inventions never came from those with 4.0 averages and all the honors and accolades; they came from average men. Einstein's GPA was 2.0 and he failed Algebra. He was turned down for admission to the University of Berlin, and went instead to the Unviversity of Switzerland at Zurich. Thank God he was a dummy! Otherwise, Adolf Hitler would have had the atomic bomb first and we'd all be saying "Sieg Heil!" If Albert Einstein were here today, and 18 years old, applying for college in the United States, he would not get in; at least not to one of the hotshot or bigshot universities. GPA and cum are asses. Consider just how many true geniuses our schools have turned away with their current policies of a stupid test based solely on memory, the GPA and cum, and no test for real intelligence exists.

Thus, the favorite saying of people who can memorize is "you're stupid."

And my reply is, "Yes, but at least I can think."

And they just don't get it.

:)


By CyberSongs at Mon, 2004/12/20 - 6:00am

Not everyone with a 2.0 GPA is Albert Einstein.


By Roberto Alsina at Mon, 2004/12/20 - 6:00am

The point is that not ever 2.0 is stupid. And that higher requirements don't necessarily equal better/smarter people.


By KT at Mon, 2008/12/01 - 6:00am

You write too much.


By ac at Mon, 2004/12/20 - 6:00am

You are talking about server side processing stuff. KHTML/Konqueror, like any other internet browser, is only responsible for rendering client side stuff so what you are asked for can at most be found in web managing software like Quanta, but not in a customer's browsing client.


By Datschge at Wed, 2004/12/08 - 6:00am

Visit this page in Firefox, and then in Konqueror. The only difference is that Firefox correctly applies the CSS stylesheet to the XML, producing a page formatted for viewing, while Konqueror does not. At least, not with version 3.3.2.

http://www.ser1.net/Files/Reviews/Sager4750.xml

--- SER


By Sean Russell at Sat, 2005/04/02 - 6:00am

Odd. Try pressing back and then forward, and please open a bug-report at bugs.kde.org


By carewolf at Sat, 2005/04/02 - 6:00am

They've already fixed the W3C page bug in Mozilla.
Check this build (for Linux):

ftp://ftp.mozilla.org/pub/mozilla/nightly/2001-09-27-06-trunk/


By scb at Thu, 2001/09/27 - 5:00am

Not only it is fixed, but now it is easier to see (it was too slow to notice before) that Konqy has some rendering problems with regard to the CSS...

Mozilla is not perfect either (it has some problems with the outer box in the "Learning CSS" part), but it renders the rounded edges beautifully (Konqy fails at that, and so does Opera).

Also, the page has a big collection of alternate style sheets (that can be accessed via View -> Use Stylesheet in Mozilla). I haven't seen a way to choose them in Konqueror. Am I missing something?


By Marcelo at Fri, 2001/09/28 - 5:00am

try this:
www.uniserv.krakow.pl
there is problem with centering email logo, and some texts in left menu frame. And most visible: Positioning background image: it should be bottom if page, but isn't.
You can view this in Explorer, Netscape and Mozilla, everything ok.


By Majki at Wed, 2001/10/10 - 5:00am

Sorry but I just validated the page and it failed. As such you are expecting the browser to work on a broken page. Sorry get the page fixed and 100% compliant then check it. Also I can at least thank a few different deities that frames are dead. xhtml 1.0 is the last version to support frames :)

I do professional web devel and I know the problems using non standard stuff causes. Even one error on the page changes the way a browser will render the page. That page is riddled with errors.


By kosh at Wed, 2001/10/10 - 5:00am

I don't want to go on about this too much, but here is the start of the MySQL code that is a problem for every Unix browser and OS:

01828
01829 MYSQL * STDCALL
01830 mysql_real_connect(MYSQL *mysql,const char *host, const char *user,
01831 const char *passwd, const char *db,
01832 uint port, const char *unix_socket,ulong client_flag)
01833 {
01834 char buff[NAME_LEN+USERNAME_LENGTH+100],charset_name_buff[16];
01835 char *end,*host_info,*charset_name;
01836 char password_hash[SCRAMBLE41_LENGTH]; /* tmp storage stage1 hash */
01837 my_socket sock;
01838 uint32 ip_addr;
01839 struct sockaddr_in sock_addr;
01840 ulong pkt_length;
01841 NET *net= &mysql->net;
01842 #ifdef __WIN__
01843 HANDLE hPipe=INVALID_HANDLE_VALUE;
01844 #endif
01845 #ifdef HAVE_SYS_UN_H
01846 struct sockaddr_un UNIXaddr;
01847 #endif
01848 init_sigpipe_variables
01849 DBUG_ENTER("mysql_real_connect");
01850 LINT_INIT(host_info);
01851
01852 DBUG_PRINT("enter",("host: %s db: %s user: %s",
01853 host ? host : "(Null)",
01854 db ? db : "(Null)",
01855 user ? user : "(Null)"));
01856

That is directly from the MySQL source. Going further in the code, when compiled it will result in can't connect to mysql socket. Find the source and tell me what you think is wrong with this code.

Posted just to back up the point about debugging MySQL source code.

css sort of becomes useless if you can't use it.


By CyberSongs at Wed, 2004/12/08 - 6:00am

The scrolling problem of Mozilla comes from the fact that the floating menu is transparent. Knowing Konqui's failure with PNG alpha channel, I think it does not render the floating menu as transparent, so he does not encounters the transparency problem. Without hardware acceleration, belnding a floating transparent menu is a CPU hog.

Not so long ago, this CSS page had a reddish opaque floating menu (instead of the blueish transparent it has now). Mozilla was fast to render it, as it is for every page on the web. The slowlyness of Mozilla and other Gecko based browsers is just due to this transparency issue IMHO. Any browser that does not implement this feature will of course avoid the problem. But it cannot be said "better".

I can't test this page with Konqui, as my linux machine is not connected to the net, but I would be really surprised if the floating menu was rendered transparent in Konqui.


By oliv at Mon, 2001/09/24 - 5:00am

It is transparent in Konqueror.


By Uwe Thiem at Mon, 2001/09/24 - 5:00am

Nice to know this.
I willl dowload the CSS and try it locally.
Gecko is quite 'old' now, but this problem with transparency is the only one I've seen so far. NS 4 and IE don't have the problem, as they don't display the floating menu, and I don't understand seeing some pages (as the one in the news) where IE is classified as excellent, when only the Mac version is trully excellent. My comment was not Mozilla/vs Konqui, but more Mozilla/ vs others (esp. the reference IE 5 and 6).


By oliv at Mon, 2001/09/24 - 5:00am

It is transparent, yes, but because the background png doesn't use
the alpha channel transparency. It just uses the "make-one-color-
completely-transparent"-thing. That's not real transparency.

By the way, does anybody know how to convince Konqueror to do some document.write()? It simply doesn't do anything when I use this method from my Javascript code.


By Benno Dielmann at Mon, 2001/09/24 - 5:00am

> By the way, does anybody know how to convince Konqueror to do some
> document.write()? It simply doesn't do anything when I use this method from
> my Javascript code.

You should probably ask me about this. :-)
See script attached, let me know if you have more questions.

Vadim


By Vadim Plessky at Tue, 2001/09/25 - 5:00am

The problem seems to be that document.write doesnt work when its inside a function, any suggestions?


By Juuso Ketonen at Mon, 2002/06/10 - 5:00am

use 'blablabla' instead of "blablabla" that worked for me


By Tyler at Wed, 2002/10/02 - 5:00am

Try a nightly build of Mozilla, or - better still - compile one yourself from CVS. The difference between 0.9.4 and the current builds is enormous in speed on this page: with a current build I can zip through the page, transparent menu and all, many times a second using the scrollbar.

Mozilla is much better in rendering this page than Konqueror (at least here, KDE 2.2.1 vs. Mozilla CVS build from a couple of days ago), as Mozilla does the transparency thing right, shows the rounded corners for the blocks, is *faster* (yes, faster, and a lot faster... That's Mozilla, mind you, I'm not even talking about Galeon of Kmeleon...), etc. Konqueror is surely getting better, but it is not on par with Gecko/Mozilla. IMnsHO...


By Frank at Tue, 2001/10/02 - 5:00am

Try a nightly build of Mozilla, or - better still - compile one yourself from CVS. The difference between 0.9.4 and the current builds is enormous in speed on this page: with a current build I can zip through the page, transparent menu and all, many times a second using the scrollbar.

Mozilla is much better in rendering this page than Konqueror (at least here, KDE 2.2.1 vs. Mozilla CVS build from a couple of days ago), as Mozilla does the transparency thing right, shows the rounded corners for the blocks, is *faster* (yes, faster, and a lot faster... That's Mozilla, mind you, I'm not even talking about Galeon of Kmeleon...), etc. Konqueror is surely getting better, but it is not on par with Gecko/Mozilla. IMnsHO...


By Frank at Tue, 2001/10/02 - 5:00am

well... it is transparent in konq, and it IS very fast to scroll that page.. at least on my PII 300.


By ealm at Mon, 2001/09/24 - 5:00am

It works fast on my Cyrix M II 200Mhz processor


By Luke at Tue, 2001/09/25 - 5:00am

seems this issue has just been reported and discussed on bugzilla
http://bugzilla.mozilla.org/show_bug.cgi?id=98252


By oliv at Mon, 2001/09/24 - 5:00am

Wtf are you talking about? Konqueror _DOES_ have transparent PNG support. YOu just cannot use transparent pngs made by Photoshop 5.0 through 5.0.3 on Windows. They are buggy. GIMP and Moz have workarounds to these buggy PNGs, but Konqueror doesn't.


By AS at Tue, 2001/09/25 - 5:00am

I'm talking about good PNG images, not buggy ones. I'm talking about the png test pages you can find on the w3c page and on libpng homepage. Most of the people I've asked on several places (as well as libpng site) do confirm that Konqueror (actually, it's more KHTML) does not properly handle PNG alpha channel.

Someone told me that it will be the case with QT 3, and when the render extension will be implemented in the video drivers (for linux and al.).


By oliv at Tue, 2001/09/25 - 5:00am

I am using konqueror HEAD (kde3), and it definately does the alpha channel on the png images on libpng.org.


By tick at Tue, 2001/10/02 - 5:00am

Pages