Readers comment on development tools SunWorld Online respondents say the biggest threat is from within By Mark Cappel

September  1996

Mail this article to a friend

The comments are reproduced as received; only some minor spelling and punctuation errors have been corrected. -- Editors

Date: Thu Aug 1 18:12:56 PDT 1996

portability is the worst problem. I want a single platform to develop to. You seem to think there's a application/scripting dichotomy. I disagree. I write *applications* using a scripting language (Python, which you didn't even include in your list :-( )

Date: Fri Aug 2 09:27:26 PDT 1996

The biggest issue we face is cross-platform development -- trying to write applications which will run on both Unix and PCs. Our latest attempt to solve this problem involved using Tcl/Tk to develop a cross-platform GUI - - what a delightful experience! Although there have been a few glitches in the recently ported PC version, in general it has given the best results of anything we have tried in the last five years. Congratulations to John Osterhout and his team at Sun.

Date: Fri Aug 2 10:04:47 PDT 1996

Application complexity and multiplatform/OS support.

Date: Fri Aug 2 11:08:23 PDT 1996

Microsoft forcing new standards no one wants or needs.

Date: Sat Aug 3 03:40:31 PDT 1996

Windows based tools lack free ware/shareware to fulfill my needs java lacks the ability to face all issues faced by onventional software like VC++. I prefer a software tool that doesnt bother me with what environment the client (MORON to be more precise) favours and java fits that need. I wish java had some facitiy to access database without making mego thro details of SQL like CDatabase and CRecordset in VC++. I wish java had low level programming faciltity like VC++.

Date: Sun Aug 4 02:53:17 PDT 1996

Let's see.... I have database tables with between 20 million to 80 million rows. The app I'm primary developer for spits out detailed analysis data of between 100GB to 200GB. Reporting pulls out and summarizes these result sets. This has to be done, in addition to other time-consuming steps, usually within a week. *EEP!* These go on Oracle (SPARC, Cray Superservers, HPUX 9/10, Sequent Dynix) or Teradata (3600, 3550). [BTW, if you're wondering, our clients are banks with $5billion+; the tables hold account data---savings, CDs, mortgages, credit cards, etc.] Data comes from various mainframes, not all consistent. I need tools that will allow me to cross platform boundaries when programming multiprocessing/multithreading apps. I can't use lock_lint, for instance, because I can't use threads (I have to use fork-model threading, since the app has to run on numerous Un*x variants, and Windows NT). I need a browser and OO design tool [SNiFF+ and Rational Rose are what we are using/going to use] with SPARCworks. I need reliable, fast ODBC and other database tools. Even with Atria ClearCase (our last SCMS, CCC Manager, blew up), we have problems with parallel development and integration. I solved most of the Un*x side problems (not the Win3.x, 95 and NT issues) Tracking of requirements, ECO, AR (anomaly reports), design specs, etc., from within a GUI, heterogenously. We use MS-Access inhouse app (E YUCH) but are switching to Scopus. I guess the bulk of it boils down to process control??

Date: Mon Aug 5 07:36:46 PDT 1996

Not Useful or Marginally Useful. Most software tools are either not useful to the seasoned developer or not integrated together well, especially in the C++ area. The graphical debuggers are great, but the code tree generators, code finders really aren't that useful to the experienced developer. I'd have to say the same for the Object Oriented Tools / Code generators like Rational Rose. They're ok for documenting large projects but after several years of work a good programmer can generate code faster and better. Complexity They are getting better Sun, HP, SGI but performance analysis is pretty complicated.

Date: Mon Aug 5 14:43:11 PDT 1996

Lack of knowledge of coworkers about types of tools available or proper use thereof. Lack of interest in a cohesive/comprehensive process and tool set. Cost Effective integrated life cycle tool set, training, etc.

Date: Mon Aug 5 22:45:59 PDT 1996

Multi-platform support. Speed, ease, and later support of development projects. Need to learn more project management and integrate it with development.

Date: Tue Aug 6 12:04:05 PDT 1996

How can I make my programmers as productive as possible with the little budget I have to spend on tools ?

Date: Wed Aug 7 06:29:30 PDT 1996

THEY ARE NOT FAST ENOUGH!!!!!

Date: Wed Aug 7 09:49:05 PDT 1996

1) Cost 2) Flexibility 3) Integration with other tools (defect database, SCM, etc) 4) Many tools are part of a "toolbox" solution to various aspects of Software Development. Therefore, buying such a tool may in fact be a commitment to a specific supplier's software development and testing philosophies and capabilities. Tools from different vendors are not really well suited to interoperate with each other!

Date: Wed Aug 7 14:21:08 PDT 1996

Fighting software license problems. Current licensing schemes create havoc when you don't fit the model of a few developers all in one location.

Date: Sun Aug 11 10:05:10 PDT 1996

Software portability and reliability.

Date: Tue Aug 13 08:38:55 PDT 1996

Getting time to develop the idea fully into a working script or program.

Date: Tue Aug 13 10:02:09 PDT 1996

Cross-platform portability ... Also portability to new OS's and backward compatability. For example, Motif/CDE is not backward compatable with Motif, yet has nice features. Motif does not have a standard C++ API. This will undoubtably cause problems in the future.

Date: Wed Aug 14 04:51:38 PDT 1996

Lack of tools for developing, debugging, and documenting largescale distributed object systems.

Date: Wed Aug 14 09:36:19 PDT 1996

The issues are cost, speed and capability. We primarily use GNU because they are free, even through they aren't the best. There is no hassle with managing licenses and getting new passwords every time we get a new machine or hire a new programmer.

Date: Thu Aug 15 10:55:04 PDT 1996

Theory is still theory. There are many nice methotds for software development, but most of them look like the invertor has newer done any software work.

Date: Wed Aug 21 20:12:48 PDT 1996

Portability is a major issue -- we endeavor to make sure all software runs on any unix box (and historically) VMS & DOS. The radio astromomy group of which I am a member is currently rewriting all the observatory antenna control and data collection software. This means we need code that will run on UNIX and a real-time operating system (pSOS) running on x86/m680xx boxes. Standard (reliable) libraries are important, most of these are public domain software

Date: Thu Aug 22 04:44:46 PDT 1996

Documentation. They mostly work, it is just a matter of finding out how.

Date: Thu Aug 22 06:17:26 PDT 1996

We are development and deployment intranets protoypes with java, cgi and databases web servers for diferent enterprises (in Colombia) that need applications for your bussines.

Date: Thu Aug 22 11:03:53 PDT 1996

Don't have time to comment here.

Date: Thu Aug 22 15:21:47 PDT 1996

We need faster development with fewer errors.

Date: Thu Aug 22 19:29:13 PDT 1996

distribute something that goodlevel for Solaris.

Date: Fri Aug 23 06:20:37 PDT 1996

Lack of cross-platform interoperability is a real problem.

SunWorld Online is published by Web Publishing Inc., an IDG Communications company, independently of Sun Microsystems Inc., which is not responsible for its contents. The opinions expressed in SunWorld Online are those of the authors or the publisher, WPI/IDG, and do not necessarily reflect the opinions of its advertisers, or of Sun Microsystems, Inc. or its affiliates.

If you have problems with this magazine, contact webmaster@sunworld.com

URL: http://www.sunworld.com/swol-09-1996/swol-09-tools.comments.html
Last update: 1 September 1996

If you have technical problems with this magazine, contact webmaster@sunworld.com

URL: http://www.sunworld.com/swol-09-1996/swol-09-tools.comments.html
Last modified: