Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

[P]
Borland Kylix -- No punches pulled.

By japhar81 in Op-Ed
Sat May 12, 2001 at 11:57:56 AM EST
Tags: Software (all tags)
Software

The Linux community has been anxiously awaiting the release of Borland Kylix, the port of their popular Delphi package to Linux. Kylix got rave reviews, won awards, and about 2 months ago, shipped. I recently got my copy, my experiences below..


Tim was nearly dancing in the halls as he came into my office with a pretty blue and white box that said Borland Kylix Server Developer on it. We had been waiting for months for Kylix to ship, and for what seemed like an eternity while purchasing finally got it for us.

Wasting no time, we got started. First order of business, build up solstice from scratch. Solstice is our reference test box, an AMD 233 with 64M RAM and a GB HD. Basically, the average box that your typical Joe would run Linux on. Yes, I have a dual gigahertz monster on my desk, but that defeats the purpose, so a 233 it is. RedHat 7.1 was our choice of OS, mostly because I like it, and, well, that's about it, I like it. The T1 we got a few days ago made the net install scream through in about an hour, we did a workstation setup and installed development and kernel development as well.

The next step, put in the Kylix CD, sit back, grin. Well, not exactly. We started the installer:

su
mount /dev/cdrom /mnt/cdrom
cd /mnt/cdrom
./setup.sh

Easy enough... It started with the first RPM and bombed with an error... oh great, is this a release or a beta? Even Microsoft software at least installs. Ok, fine, we'll cheat. We went in and installed the 6 RPMs by hand. That was a no-no. The RPMs dumped all the .so's in to the /bin instead of /lib directory. Ok, we can move those. While we were doing this, we noticed some odd libraries, bplkernel32.so, bplgdi32.so, etc. My initial thought was, great, another port of win32 libraries to Linux, and they call it native. I reserve judgment on this, maybe its so I can recompile on windows later, I don't know. The wine-related libraries are also somewhat suspicious to me, but oh well...

Tim suggested we try to install Kylix as a user instead of as root, this worked, but it didn't do the RPM install, don't ask me why. Their installer needs work, big time. Finally, we got this silly thing installed, and fired it up. It was a bit slow, but it's a slow machine, its forgivable. Next, I opened the clock demo, it came up, it looked cool, I was happy again. Well, not quite. I clicked the run icon, and it ran. Then I closed the window, and Kylix was frozen. And I mean frozen. We had to kill -9 the thing.

This couldn't be right, we expected a lot better from Borland, and we always got it, until now... Checking Borland's site, we found no patches for Kylix... We decided that maybe something in RedHat should be updated. I use Ximian GNOME anyway, and Red-Carpet did the trick for updating the box to the latest everything.

The updates went through without problem, but I did't see my patching up2date, rhn_register, and netscape doing anything to fix Kylix...

The final test before I fdisked the thing was to write a simple little app and see how well it worked. Basically, a text label and a button. Click the button, get `Hello, World' on the text label. Piece of cake.

Thirty seconds flat, and it all worked. Back to being all happy, for the time being... The next test was to mess around with some demos. All of a sudden, I see windows-looking abort - retry - ignore dialogs! What is this? I use Linux for a reason, I DON'T WANT TO USE WINDOWS. Come on Borland, get with it. Opening the textBrowser demo Kylix blew up on me again, kill -9, AGAIN.

I expected Kylix to have bugs. It's a 1.oh product from a windows shop that's just crossing over to Linux. Its understandable, its expected, its not uncommon. The level of problems with Kylix is just awful though, far worse than Id expect from a reputable shop like Borland. When it comes to Kylix though, I think it's best to wait for SP1, or better yet, the 6.0 release of Delphi for win32 and Linux. Flame me if you will, I totally agree that Kylix is a step in the right direction, but with products shipping in this state and being called 'enterprise-level commercial-quality software', windows really has nothing to worry about, because with this many bugs, I don't dare use Kylix on a real test box, much less on one of my production machines.

Sponsors

Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure

Login

Poll
Is this review totally off-base?
o Yes 6%
o No 14%
o I dont program in Pascal 34%
o Kylix? Whats Kylix? 14%
o Haven't tried it yet. 29%

Votes: 75
Results | Other Polls

Related Links
o Also by japhar81


Display: Sort:
Borland Kylix -- No punches pulled. | 35 comments (35 topical, editorial, 0 hidden)
Why is it... (3.22 / 9) (#1)
by trhurler on Fri May 11, 2001 at 05:12:02 PM EST

that the same sort of people who like Delphi also like RedHat? I think there is something vaguely epistemological regarding one's choice between substance and style to be had here, but I can't quite put my finger on it:)

--
'God dammit, your posts make me hard.' --LilDebbie

Put down the bong (3.40 / 5) (#3)
by shoeboy on Fri May 11, 2001 at 05:49:26 PM EST

And back away from the keyboard.

the same sort of people who like Delphi also like RedHat? I think there is something vaguely epistemological regarding one's choice between substance and style to be had here, but I can't quite put my finger on it:)

You seem to be inferring that Red Hat and Delphi have style, but not substance.

If you were sober you'd realize that they have neither. Sorry trhurler, but you're batting 500 on this one.

--Shoeboy
No more trolls!
[ Parent ]
Nah (4.60 / 5) (#6)
by trhurler on Fri May 11, 2001 at 06:03:59 PM EST

I think the whole thing fits perfectly; I didn't actually say they HAD style. Usually, if you're chasing style over substance, you get neither.

--
'God dammit, your posts make me hard.' --LilDebbie

[ Parent ]
switching to school marm mode (4.00 / 2) (#13)
by pookieballs on Fri May 11, 2001 at 11:43:22 PM EST

You're misusing the word epistemological. Epistemology is the branch of philosophy dealing with the nature of knowledge and knowing, and the validity of knowledge.

[ Parent ]
You're half right (none / 0) (#26)
by trhurler on Mon May 14, 2001 at 11:22:02 AM EST

Your definition of epistemology is essentially correct. In fact, I used it in this manner; there is a fundamental difference in the epistemologies of those who seek out effective methods of dealing with the world around them and those who seek out ways to look cool while dealing with the world, and it was specifically at that level that I targeted my rather tongue-in-cheek comment. I could have made it clearer without changing the meaning TOO much by substituting metaphysical, as the two variants on epistemology have corresponding differences in their basic view of the world they inhabit and their relationship to it, but to most people, metaphysics doesn't mean anything anymore; the word is too overused. It is entirely possible that I spend too much time on such subjects; I have noticed that most people do not actually make jokes related to them:)

--
'God dammit, your posts make me hard.' --LilDebbie

[ Parent ]
I think a better term would be teleology (none / 0) (#29)
by pookieballs on Mon May 14, 2001 at 12:29:51 PM EST

As what you appear to be pointing out is less a matter of knowing, and more a matter of motivating "force" behind a phenomenon.
This, of course, is an outmoded concept (it essentially died out with the scholastics, and is merely dragged along for show through the 17th and 18th centuries) particulary in light of some 20th century existentialism; to paraphrase Sartre, the phenomenon only indicates itself and the series of phenomena of which it is a part, there is no reality "behind" the appearance. Thus, the apparent preferences of a category of system user indicates nothing but itself and gives no suggestion of purpose, nor any noumenal world beyond what is percieved. This, naturally, has some disastrous consequences for the nature of metaphysics as you would appear to concieve of it. Esse es percipens, as it were, since the perciever is now firmly that which confers significance on phenomena.

At least I think. I smoked a lot of ganja in college.

[ Parent ]
Not quite (none / 0) (#30)
by trhurler on Mon May 14, 2001 at 12:38:53 PM EST

Teleology has a broader meaning than you ascribe to it, but in no sense does it describe what I meant. As an example, I am going to buy a car. A totally impractical little two seater with no cargo room to speak of. I go out and look at performance, pricing, reliability, and so on. I don't just take the company's word. Why? Because I believe this will lead to a superior selection. What do most people do? They decide which one looks cool and then figure out a way to finance it come hell or high water. If it breaks down, they don't care; they pretend this is a fluke. If it doesn't perform, they don't mind, because they bought it to look cool. They don't know a sports car from a cell phone, and they don't want to. They want to look like they do.

The key is this: they don't just want to look like they're in the know rather than be in the know because they think it is easier - to them, the former IS the latter. They honestly believe they "know" sports cars. After all, isn't this a shiny, sleek looking machine?!

Of course, I'm omitting lots of details, like why they make this mistake, how they came to such a state, and so on, but that's a long, involved mess and I doubt I have all the answers anyway. The key insight is that, for whatever reason, these people think style IS substance - a matter of belief, not some "underlying motive force."

By the way, Sartre was a headcase. He realized he was responsible for his own life and decided he didn't like that, and so he wrote books about it. Then some idiots made the mistake of taking him seriously.

--
'God dammit, your posts make me hard.' --LilDebbie

[ Parent ]
This does not sound good. (2.25 / 4) (#2)
by WinPimp2K on Fri May 11, 2001 at 05:13:45 PM EST

And I did want to see Borland do something spectacular. Looks like I'll be stuck with MSFT products for a while longer.

Damn

Why? (3.00 / 1) (#22)
by S5o on Sat May 12, 2001 at 09:17:57 PM EST

Microsoft products for Linux?
Kylix is a (apparently rather buggy, although I had none of the problems described in this article) port of Delphi to Linux. If you're doing Windows devlopment, Delphi is a superb product. (Except coreide.bpl screwing up ever few thousand lines.)

[ Parent ]
Kylix, Killed by Borland before its time? (3.25 / 4) (#4)
by cable on Fri May 11, 2001 at 05:52:32 PM EST

Have they gotten the Free version out yet? If not, until they do the Students and Free Software people will be having a hard time getting a copy and learning/supporting it.

A 1.0 product is supposed to have bugs, but you'd think they have an update by now? Even Microsoft would have had a critical update by now if it was their product. I think Borland is going to the dogs, unless they can pull a rabbit out of the hat, I think they are doomed. Maybe Klyix 2.0 will be better?

Anyone remember learning with Turbo Pascal 2.0 or 3.0? I do. It may be better to download them from Borland Museum and then run the DOS version under DOSEMU than use the buggy Kylix you reviewed. Maybe someday they will put Delphi 1.0 or 2.0 in the Museum?

Either that or download Virtual Pascal for Linux a Turbo Pascal clone.

------------------
Only you, can help prevent Neb Rage!

Bugs (4.66 / 9) (#5)
by wiredog on Fri May 11, 2001 at 06:03:04 PM EST

A 1.0 product is supposed to have bugs

Man, you've really been brainwashed by the commercial vendors. An 0.x release is expected to have bugs. A 1.0 release, however, is supposed to not have bugs. While any non-trivial software will have bugs, gold code is supposed to be free of serious, show-stopping bugs. The bugs in a 1.0 release should be small, hard to find, and unnoticed by the average user.

The idea of a global village is wrong, it's more like a gazillion pub bars.
Phage
[ Parent ]

If bugs == 0 then bugs = 1 else bugs = bugs + 1 (none / 0) (#31)
by cable on Mon May 14, 2001 at 03:02:23 PM EST

I have never found a product to be completely bug-free. Just that companies being very good at hiding bugs as "undocumented features to be changed in future releases". Crashing the whole product is one very large "undocumented feature" indeed! :)

------------------
Only you, can help prevent Neb Rage!
[ Parent ]
Bugs (4.00 / 4) (#8)
by ell7 on Fri May 11, 2001 at 07:00:01 PM EST

I guess a few bugs in a 1.0 product are forgivable, if they can be worked around. Showstoppers are huge no-no, though, I mean come on!

Nothing is more annoying than paying for software and then having it not work! Having a non-beta product lock solid after a few minutes of use is utterly unforgivable.

[ Parent ]
dot ohs and bugs (4.50 / 2) (#18)
by zenspider on Sat May 12, 2001 at 01:55:23 PM EST

A 1.0 product is supposed to have bugs...

No, a dot-oh product is NOT supposed to have bugs. It will most likely have bugs, but it isn't supposed to. This type of attitude is what allows companies to put out shit for a dot-oh and it is happening more and more often. We need to hold them (and us) to a higher standard.

I'd personally rather have less features and more stability per release with a faster release cycle. Too many companies put out huge releases once a year (or more). Most of that time is spent in development of the forty-thousandth-feature-request-from-marketing. Most of the time should be spent spit between design and QA. Coding isn't hard. Making a usable and good product is.

[ Parent ]

Clarification (none / 0) (#23)
by japhar81 on Sun May 13, 2001 at 01:20:40 PM EST

When I say a .oh should have bugs, I mean bugs, like oops, that icon didnt refresh. I dont mean huge, gaping mistakes like the whole damn thing locking. Thats not a bug, that goes far beyond what I would consider a bug in my software, and I damn well wouldnt let it get out the door of one of my commercial products. Or one of my OSS projects either to be quite frank...

<H6>Rome is always burning, and the younger generation never respects its elders. The time of your second coming, japhar81, is no exception. -- Aphasia</H6&gt
[ Parent ]
What's your point? (3.41 / 12) (#7)
by regeya on Fri May 11, 2001 at 06:45:34 PM EST

All of a sudden, I see windows-looking abort - retry - ignore dialogs! What is this? I use Linux for a reason, I DON'T WANT TO USE WINDOWS. Come on Borland, get with it.

So what do you want? An xterm to pop up with the message "Segmentation fault (core dumped)"? Dialogs with buttons labeled "Dismiss"? Apps that die for no apparent reason? OpenView? *shudder*

So, you're saying that because it's a product for Linux, that the interface should be totally overhauled? What do you prefer?

Bah, go play with your 50 xterms and keep your review to yourself.

[ yokelpunk | kuro5hin diary ]

or.. (2.85 / 7) (#9)
by rebelcool on Fri May 11, 2001 at 08:18:07 PM EST

as most linux interfaces seem to have done, implement a shoddy half-assed version of things windows and mac were doing several years ago.

COG. Build your own community. Free, easy, powerful. Demo site
[ Parent ]

and perhaps.. (3.00 / 3) (#10)
by slaytanic killer on Fri May 11, 2001 at 09:23:21 PM EST

it would be really pathetic, since the two are grimy works built on the bones of Amiga and GeoWorks.

[ Parent ]
which.. (3.00 / 3) (#11)
by rebelcool on Fri May 11, 2001 at 09:54:11 PM EST

in the end comes from some clever people at xerox.

COG. Build your own community. Free, easy, powerful. Demo site
[ Parent ]

and... (4.00 / 4) (#12)
by Delirium on Fri May 11, 2001 at 10:06:47 PM EST

Who stole all their ideas from me.

[ Parent ]
Douglass Engalbart (2.50 / 4) (#21)
by delmoi on Sat May 12, 2001 at 03:27:41 PM EST

Who want everyone to use his stuff anyway.
--
"'argumentation' is not a word, idiot." -- thelizman
[ Parent ]
What the hell? (3.00 / 1) (#32)
by delmoi on Tue May 15, 2001 at 02:36:26 AM EST

Why was this rated a '1' I mean, was it spam? was it a personal attack?
--
"'argumentation' is not a word, idiot." -- thelizman
[ Parent ]
Has Borland had it? (4.16 / 6) (#14)
by cezarg on Fri May 11, 2001 at 11:44:46 PM EST

Your review is frightening. It frightens me because it's so so familiar. Except in my case the product is called JBuilder and it's for Java. The common factor: JBuilder is also made by Borland. JBuilder is a mess. No not a mess, it's a huge, unstable, fucking unusable, ugly, buggy piece of crap that I wouldn't release as shareware let alone charge thousands of dollars for it. Borland is a sham and their only reason for existance is that they have no viable competition in the Pascal and Java worlds. They truly deserve to go bankrupt. I don't care how good Turbo Pascal 6 was. Ever since the days of windows this company has had a stream of buggy unstable products that just failed to perform. Their UI is designed by patients of mental institutions and their QA team clearly sabotages the company. JBuilder bugs are not some tiny wee obscure things that very rarely bite. They are flipping huge stoopid errors that should have been fixed in version 0.3 let alone slip into 1.0. Example: Adding a new library to a JBuilder project will only get picked up by the runtime in the debugging mode only after you restart JBuilder altogether. This is just not acceptable. Such bugs I wouldn't tolerate in any product. It's so damn obvious that there are some serious design issues within JBuilder itself if bugs such as this one are left outstanding. There is a whole slew of problems with JBuilder that are in a similar vein. I could go on but I'm digressing too much already. Borland is desperate to put something usable out the door and frankly it's just not happening. The only decent piece of software that they offer is Visibroker and the reason it's good is because it was coded by programmers from Visigenic not Borland. I hope Borland go bust. This industry needs competition but we need real competition and not some shoddy substandard overpriced garbage that Borland makes.

re: Has Borland had it? (none / 0) (#27)
by technik on Mon May 14, 2001 at 12:02:17 PM EST

Regarding JBuilder I must ask,"What version are you using?" and "What are you doing with it?"

I've suffered since 2.0 and JBuilder has improved in each successive release. JB4.0 is as stable as VA for Java and more stable than NetBeans. I use it daily under Win32 and Linux switching between IBM and SUN JDKs on the same OS. I can recommend JB4. That was not the case with 2.x or 3.x.

That said, I probably don't use half of the features. I write mostly servlets or console-based services and could just as easily switch to using Vim (except that I would miss the class browser and method completion but that's what ctags is for, right?). I've done only two or three small GUI apps and I don't use the "wizards" at all. I like the editor focus of the user interface and that there are keyboard shortcuts for just about everything. I also rely on a handful of OpenTools (Junit, JRefactory). I find netbeans, even 3.2, to be completely idiosyncratic and buggy and VAJ is huge and too tightly structured for use with WebSphere to be good alone. JBuilder is a good compromise.

I often wonder what focus, if any, Borland has but the products have been good.
- technik

[ Parent ]
JBuilder 4 indeed. (none / 0) (#28)
by cezarg on Mon May 14, 2001 at 12:14:30 PM EST

Just try using the builtin debugger. There is a whole slew of bugs (including the one mentioned in the original post). Their class browser is only marginally useful because it parses your stuff in realtime and bails out on the first error you make sometimes you don't get to see what's in your class. It bit me several times (I was thinking that my scc f**ked up my source files). About the only thing I like about JBuilider is that I can rename my files within the IDE. Everything else seems inferior to the now defunct J++. J++ may have been a java bastardisation but it ruled as an IDE. I've started playing with Forte and I think I like it quite a bit more than JBuilder. It definitely feels more solid.

[ Parent ]
Aye, it is sad (4.00 / 2) (#15)
by Alhazred on Sat May 12, 2001 at 08:14:25 AM EST

I had the same problems with JBuilder on my Mandrake box as one of the earlier posters. Haven't tried Kylix yet, but I am not really all that inspired to do so by what I've seen so far.

JBuilder on Linux is a flaming you-know-what. 3.5 was totally unusuable in any way shape or form. The installer was ghastly and didn't function at all. Once I DID get it sort of installed I had to deal with license key crap for a week just to get it to run. At that point the entire product was simply so buggy in so many 100's of ways it was unuseable.

This is sad. Delphi was and is a kick-ass development environment on win32. 100 times better than VB or VC++. I was hoping these guys would do the same for Linux, but I guess we will have to wait for V2 of Kylix and see what happens. Maybe they will get a clue, but frankly I think it might be best to just go with python and Activestate's stuff for now. At least it works and does some usefull stuff.
That is not dead which may eternal lie And with strange aeons death itself may die.
One problem (1.40 / 5) (#16)
by darthaya on Sat May 12, 2001 at 10:18:06 AM EST

Linux has way too many different distributions. And you can't expect to slap the system on any of them with even different c libraries to work right out of box.

That is one good thing about windows, you don't have to write it once, and test everytime you run it. :-)

I disagree (5.00 / 1) (#17)
by tzanger on Sat May 12, 2001 at 12:33:01 PM EST

I disagree wtih you on a very basic level.

Linux has way too many different distributions. And you can't expect to slap the system on any of them with even different c libraries to work right out of box.

That's so untrue it hurts.

You clearly state that "requires glibc2.2, ncurses 5.x and kde2" and you won't have trouble. If they don't have the minimum requirements, it won't run. Same as in Windows.

No, better than in Windows. I've never had trouble running software "designed" for RedHat or Debian unless they use their "proprietary" extensions to a base Linux system.



[ Parent ]
Yep! (none / 0) (#25)
by jynx on Mon May 14, 2001 at 06:00:18 AM EST

Companies really should be wise to this. Surely Borland could spare a couple of people for a couple of weeks to test everything on some of the popular distros.

--

[ Parent ]

Yeh, right... (3.00 / 1) (#20)
by delmoi on Sat May 12, 2001 at 03:24:16 PM EST

That is one good thing about windows, you don't have to write it once, and test every time you run it. :-)

Well, maybe not to the same extent as Linux, but there are some pretty big incompatibilities between various win32 distributions. A lot of stuff programmed for 98 doesn't work on NT due to people not 'following the rules' correctly, and the 9x versions are missing a lot of functionality, most notably the Unicode libraries.
--
"'argumentation' is not a word, idiot." -- thelizman
[ Parent ]
RAD GUI for Linux. (3.66 / 3) (#19)
by tkatchev on Sat May 12, 2001 at 02:40:52 PM EST

May I suggest Glade + Python for rapid GUI development for Linux? I've tried it a little bit (I do mean a _little_) and from the limited experience the thing is absolutely sweet. Any knowledgeable experience here? Thanks.

   -- Signed, Lev Andropoff, cosmonaut.

Hmm... This didn't happen to me. (4.33 / 3) (#24)
by ghjm on Sun May 13, 2001 at 11:49:13 PM EST

I had a couple install glitches similar to what was described - I was installing on Red Hat 7.0, and Kylix wants to upgrade various things, which broke various other things, etc...so I agree, the installer definitely needs a lot of work. However, once installed, I found the product to be stable, at least.

My problem with it is simply that it doesn't do what I currently want. If I was interested in developing GUI client-side apps on Linux, I'm sure Kylix would float my boat. However, that's not what I want. I have a Win32 MIDAS middle tier, built using remote data modules; I want to recompile it for Linux, then connect to it with my existing Win32 workstations using the Delphi 5 / MIDAS 3 TSocketConnection.

Problem is, Kylix Server Developer doesn't include MIDAS support. (!)

So if I want to take my existing all-Win32 Delphi/Delphi/Interbase-NT three-tiered application and port it to Delphi/Kylix/Interbase-Linux, which seems like a pretty obvious thing to want to do, I don't have a transport protocol. The best I can figure is to use the Delphi/Kylix dataset's SaveToStream method to export XML packets, then write my own transport to get them between the client and the middle tier...but I really didn't want to have to rewrite the MIDAS transport layer in order to get this to work.

This was the major benefit I thought Kylix would bring to the table, and it showed up without it. All I can hope is that D6 provides something, but more likely, D6 will have some new transport protocol which *will* be compatible with Kylix, but not until Kylix 2.0, maybe in late 2002. In the meantime, without MIDAS, I'm left wondering what exactly Kylix has to offer that Free Pascal doesn't already give you? I feel like my life has been on hold waiting for Borland since about 1997.

Also, I think the whole Kylix / Linux / Open Source thing from Borland was just Dale Fuller's idea of a smart stock play back when Red Hat and VA Linux were looking stratospheric...else how do you explain the Interbase debacle? Now that Linux stocks are in the toilet, what do you want to bet Kylix gets starved for resources? I'd be amazed if a Kylix 2.0 ever makes it out the door. I'm not sure who's the worse asshole, Dale or the idiot board of directors he reports to, but their motivations are clear in this regard. Kylix was just their attempt to play buzzword bingo, and they picked the wrong buzzword.

The best thing that could happen to Delphi/Kylix would be for Borland to just give it up and declare bankrupcy, and sell their products to (say) CA...who would as usual pursue a minimal-investment, cash-cow strategy but would at least provide competent support to the existing customer base. No, I'm not saying this would be a good thing, just that it's the best we can hope for, and that it would be better than leaving those idiots at Borland management in charge...

This whole situation reminds me of when VB came out and TPW just wasn't cutting it any more, there were two or three dark years where all us Turbo folks had to go be VB developers before Delphi came out. Why can't Borland just get it together for once and all? Why don't they realize that their customers will keep them in business, regardless of their stock price, if only they'll give us just the slightest taste of the stuff we actually need?

SIGH.


Do your homework (4.66 / 3) (#33)
by pcisar on Tue May 15, 2001 at 03:23:25 AM EST

Well, you're tested Kylix and it blow you in face, but you had to do your homework first. Read the readme, for example, or look at web or Borland newsgroups.
  1. RH 7.1 is not in list of supproted distros (RH 6.2 and 7.0 are). Of course, Kylix should work on it, but RPM manager in early RH7.1 was a bug when works on relocatable rpm's. Just visit Borland's public newsgroups (there are +15 just about Kylix) and ask there. You'll get prompt answer.
  2. Older versions of glibc have bugs (discovered by Borland). Borland submit patches to the community and main Linux distributors. Old distros have to be patched with packages provided on Kylix CD.
  3. Kylix IDE (like Delphi) is written in Object Pascal and with use of VCL component library. For Kylix, the VCL was replaced by CLX multiplatform library based on Qt. In first version of Kylix, the IDE is a mix of VCL and CLX code. Because VCL binds itself to Win32 API, the Kylix IDE was linked against Winelib.
  4. Kylix use CLX library based on Qt, so it works seomewhat better in KDE, especially on underpowered hardware you're used (hint: GTK+ and Qt libraries are in memory when you're use GNOME and Kylix IDE is also memory-hungry).
  5. Some demos are "dangerous" (not properly ported from Delphi or betaversions of Kylix).
  6. There are no patches for Kylix yet, but you should understand that almost all people from Kylix team now works hard on cross-platform parts in Delphi 6 (the second piece of Borland's cross-platform solution). It's supposed to be out in a month or so. After that, you can expect update to Kylix when changes made to CLX for D6 would be merged back to Kylix.
  7. There is a FreeCLX project at SourceForge created and maintained by Borland people with full source of GPL'ed CLX (in CVS). There're some patches to CLX already.
Of course, Kylix 1.0 have bugs, but otherwise it's a damn good piece of software, and it's rock-solid when tuned to your working environment. I'm using it on my Mandrake 8.0 (and some time before on 7.2) with KDE 2.x regulary for heavy development and I'm very satisfied.

Howgh!

Borland Kylix - Give us a break, mate (4.00 / 1) (#34)
by MrApple on Tue May 15, 2001 at 08:12:19 AM EST

This guy didn't bother to read the Readme which would have told him what he needed to know about pre-installation setup and how different privileges settings would affect his installation - for example, how to test his setup and how to apply the supplied Pre-install patches; how running the Install as root would install the packages whereas installing as user wouldn't. And quite a few other important details about installing.

He also could have discovered that the release wasn't certified for RH 7.1. All he's describing is a classic admission of "I didn't RTFM".

I, too, had problems with the installation. I was ready to blame anybody, including the Borland documenters for not anticipating my problem!. Actually, I was ready to disembowel myself. I was at RedHat 6.2 reinstall number 4 before I discovered the problem - it was hardware and OS-related and it had absolutely nothing to do with Kylix.

Once that was fixed the install went like clockwork. If I wasn't such a dummy about Linux, I wouldn't have had the problem in the first place.

I have Kylix Developer Edition installed on a pretty ancient IBM 686 with a clock speed of 200MHz. The Gnome interface doesn't work properly, as I expected because of a missing Gnome Enlightenment update (it doesn't freeze, either) but it goes like a bomb in KDE = "works as expected".

I think Kylix R & D have done a fantastic job, considering that the whole Kylix/D6 development project is cross-platform. Kylix is as good as the first release of Delphi was. It deserves better than to be reviewed by some dilettante who thinks all you have to do to install a complex new IDE software on ???Linux??? is hit the Go button.

/os

[ Parent ]

my experience... (none / 0) (#35)
by heph on Tue May 15, 2001 at 03:59:49 PM EST

could not have been more different.
I got my copy today and so far it has worked perfectly.
I installed it on my redhat 7.1 machine (not as root to avoid the rpm bug in 7.1) and managed to run a few test programs without any problems.
It also installed without any problem as root on mandrake 8.
All in all I have been really impressed with kylix so far, and I really hope borland stick with it.

Borland Kylix -- No punches pulled. | 35 comments (35 topical, 0 editorial, 0 hidden)
Display: Sort:

kuro5hin.org

[XML]
All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
See our legalese page for copyright policies. Please also read our Privacy Policy.
Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
Need some help? Email help@kuro5hin.org.
My heart's the long stairs.

Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!