Program-source in ISAM-files?

2008-06-06 13:03:10 +0200

Hi there,

since a few months, I have been given the job of maintaining
the PL1-compiler on BS2000. This is a very old product, rooted
back into the late 70-ties, and even on another platform at that
time ( BS3 / Telefunken TR440). Once upon a time this
PL1-project was a busy one, with many team members - but now,
in it's old age (and my old age), it's only me who's doing
maintenance.

I have experience in a few other compiler project on BS2000
- mainly Ada, COBOL, and an ESQL-Preprocessor for
COBOL. Among these, COBOL is also a very old project.

What I am uneasy with is that in such old projects, Source
Code is often in ISAM-files. That makes it difficult to edit
source with other editors than on the mainframe - and
mainframes are not known for efficient editors. Second,
comments in the code often refer to line numbers (ISAM-keys)
rather than functions or modules. Third, modern
Source Management Systems cannot be used for ISAM-files,
rather old legacy source code management systems on that
mainframe must be used - and often enough, each project has
it's own source management system.

What I want to know is: How common is that situation (source
code in ISAM files)? As for me, I know no other mainframe
system than BS2000, so I cannot tell. Are there hidden benefits
of ISAM-files for Source-Code which I do not know?

As for PL1, I have moved the code to a Unix-based CVS two
month ago - and it feels like breathing more freely.

Herwig

Herwig Huener wrote:
> What I want to know is: How common is that situation (source
> code in ISAM files)? As for me, I know no other mainframe
> system than BS2000, so I cannot tell. Are there hidden benefits
> of ISAM-files for Source-Code which I do not know?

Yes. On time-sharing systems 40 years ago, it cost too much in RAM to
hold entire source files in memory while they were being edited.

No, I'm not joking.


--
John W. Kennedy
"But now is a new thing which is very old--
that the rich make themselves richer and not poorer,
which is the true Gospel, for the poor's sake."
-- Charles Williams. "Judgement at Chelmsford"

In <g2b5k0$aea$1@nntp.fujitsu-siemens.com>, on 06/06/2008
at 01:04 PM, "Herwig Huener" <Herwig.Huener@fujitsu-siemens.com> said:

What I am uneasy with is that in such old projects, Source
Code is often in ISAM-files. That makes it difficult to edit
source with other editors than on the mainframe - and
mainframes are not known for efficient editors.

I'd have to disagree with both. I have had no difficulting with using a PC
editor for mainframe files with line numbers, and some mainframe editors
are quite efficient. Don't assume that BS2000 is representative of the
mainframe world.

>Second, comments in the code often refer to line numbers (ISAM-keys)
>rather than functions or modules.

Ouch!

>Third, modern Source Management Systems cannot be used for
>ISAM-files,

Actually, they can, but there's not much point to it.

>rather old legacy source code management systems on that
>mainframe must be used

Only if you want to. And "legacy" means "it works" ;-)

>and often enough, each project has
>it's own source management system.

Ouch!

>What I want to know is: How common is that situation (source code in ISAM
>files)?

Source code in ISAM files may be common, but comments referring to line
numbers are insane. Please tell me you're joking.

--
Shmuel (Seymour J.) Metz, SysProg and JOAT <http://patriot.net/~shmuel>

Unsolicited bulk E-mail subject to legal action. I reserve the
right to publicly post or ridicule any abusive E-mail. Reply to
domain Patriot dot net user shmuel+news to contact me. Do not
reply to spamtrap@library.lspace.org

Herwig Huener wrote:
> 2008-06-06 13:03:10 +0200
>
> Hi there,
>
> since a few months, I have been given the job of maintaining
> the PL1-compiler on BS2000.

Congratulations! Welcome to the family!
....

>
> What I want to know is: How common is that situation (source
> code in ISAM files)? As for me, I know no other mainframe
> system than BS2000, so I cannot tell. Are there hidden benefits
> of ISAM-files for Source-Code which I do not know?

When you say "ISAM" I assume you're talking about a system that
maintains a table of pointers to records, hopefully not what IBM call's
ISAM. I believe this was a fairly common system. The CMS filesystem is
done this way too, as far as I know.

It has a few advantages.
1) In the days of limited memory it was simple to edit files larger
than could fit in memory. This is obviously in the past. Nowadays an
editor just reads an entire file, edits it, and writes the whole thing
back out.
2) There is no restriction on the characters you can store in a file.
With a unix (or peecee) type file system Linefeed, and maybe
carriage-return are reserved and can't be data. NULs are also problematic.
3) The statement number can be assumed to be invariant, ubless you
resequence the file. Much older non-interactive editors would allow
insert, delete, and replace by statement number. Usually files had
statement numbers increment by 10 or 100 to allow lots of inserts.
Statement numbers can still be handy at times, but probably only if
you're used to working with them.

I would expect that "modern" editors, as they read the file into memory,
build the same type of index on the fly.

>
> As for PL1, I have moved the code to a Unix-based CVS two
> month ago - and it feels like breathing more freely.
>

Probably a good move. Your unix sysadmins are no doubt making sure that
everything is being backed up regularly. Did you try a restore (to
another machine or directory) to make sure you're getting usable backups?

Best of luck with your project.

> Herwig
>
>

"Herwig Huener" <Herwig.Huener@fujitsu-siemens.com> wrote in message
news:g2b5k0$aea$1@nntp.fujitsu-siemens.com...
> 2008-06-06 13:03:10 +0200
>
> Hi there,
>
> since a few months, I have been given the job of maintaining
> the PL1-compiler on BS2000.

Who made the BS2000 ?

20 Prairial an 216 de la Révolution
10:10:10 +0200

Peter Flass wrote:

> ...
>
> Congratulations! Welcome to the family!

:-)

> ...

> 2) There is no restriction on the characters you can store in a file.
> With a unix (or peecee) type file system Linefeed, and maybe
> carriage-return are reserved and can't be data. NULs are also problematic.

Moving files between a mainframe and a PC or
Unix is always a source of entertainment. When
I retire, I will, living with my
Linux-monoculture at home, miss all that!

> ...

>> As for PL1, I have moved the code to a Unix-based CVS two
>> month ago - and it feels like breathing more freely.
>>
>
> Probably a good move. Your unix sysadmins are no doubt making sure that
> everything is being backed up regularly. Did you try a restore (to
> another machine or directory) to make sure you're getting usable backups?

My Unix SysAdmin reads my very thoughts. It's myself,
by the way.

PL1 is not the first project I moved into a CVS.
Also, for the sake of safety, I triple the efforts:
There are *three* CVS repositories: One local
in CygWin, one on Solaris, and one on Reliant-Unix.
Every commit is mirrored on all three machines - on
Unix, you can script that easily.

Backup: Self-understood. Also, I make cryptographic
hashes from each backup, so not a single bit will
die unnoticed.

BTW: Today, I would prefer subversion as a software
management system - but you cannot change a sowftware
management system every other day, can't you?

Herwig

--
**************************************************
* Herwig Huener http://www.quantenrente.de *
* Assistierender QuantenScharfRichter *
**************************************************

20 Prairial an 216 de la Révolution
10:25:00 +0200

robin wrote:

> Who made the BS2000 ?

Siemens. And BS2000 is running Europe.

Herwig

--
**************************************************
* Herwig Huener http://www.quantenrente.de *
* Assistierender QuantenScharfRichter *
**************************************************

20 Prairial an 216 de la Révolution
10:38:00 +0200

Shmuel (Seymour J.) Metz wrote:

> ...

>>rather old legacy source code management systems on that
>>mainframe must be used

> Only if you want to. And "legacy" means "it works" ;-)

Yes, those systems I know about work - but they are
hard to use. And if the last maintainer retires,
they better continue to work, because nobody can
fix problems.

> Source code in ISAM files may be common, but comments referring to line
> numbers are insane. Please tell me you're joking.

I'm not. Actually, there are people who cannot
programm without their line numbers. It's alien
to me and probably to most - but to these people
the idea of refering to line numbers comes
easily. I think, it's a matter of live-long
practice.

What I consider harmful is if such way of working
is forced upon others who are used to think in
different ways.

That's one real danger: a young innovative mind
joins an old old legacy project and is forced to use
stoneage methods. This can crush a professional
career easily.

However, if such young innovative mind is the only
one left is such a project, hope is rising again,
because he brings with him a number of new weapons:
CVS, emacs, lex, yacc, awk, cscope, to name a few.

Mmh, cscope - I would like an pl1scope ...

Herwig

--
**************************************************
* Herwig Huener http://www.quantenrente.de *
* Assistierender QuantenScharfRichter *
**************************************************

Herwig Huener AQSR wrote:
>>
>> Probably a good move. Your unix sysadmins are no doubt making sure
>> that everything is being backed up regularly. Did you try a restore
>> (to another machine or directory) to make sure you're getting usable
>> backups?
>
>
> My Unix SysAdmin reads my very thoughts. It's myself,
> by the way.
>
> PL1 is not the first project I moved into a CVS.
> Also, for the sake of safety, I triple the efforts:
> There are *three* CVS repositories: One local
> in CygWin, one on Solaris, and one on Reliant-Unix.
> Every commit is mirrored on all three machines - on
> Unix, you can script that easily.
>
> Backup: Self-understood. Also, I make cryptographic
> hashes from each backup, so not a single bit will
> die unnoticed.

Sorry if I sounded a bit patronizing. I wasn't sure of your level of
experience. Mainframe people take backups and such for granted, and
often assume the same for smaller machines, but this is sometimes not
the case. I'm paranoid about backups.

Herwig Huener AQSR wrote:
>
> That's one real danger: a young innovative mind
> joins an old old legacy project and is forced to use
> stoneage methods. This can crush a professional
> career easily.
>
> However, if such young innovative mind is the only
> one left is such a project, hope is rising again,
> because he brings with him a number of new weapons:
> CVS, emacs, lex, yacc, awk, cscope, to name a few.

This is true, but sometimes the youngster goes too far and throws out a
system that works in favor of something new built with tools s/he
understands. It sounds like you have the balance about right. By the
way, what language is your compiler written in, if that's not
"confidential"?